Counter-argument to Searle's Chinese Room argument
Anssi Hyytiainen
---
Imagine a native English speaker who knows no Chinese locked in a room
full of boxes of Chinese symbols together with a book of instructions
for manipulating the symbols. Imagine that people outside the room send
in other Chinese symbols which, unknown to the person in the room, are
questions in Chinese. And imagine that by following the instructions in
the program the man in the room is able to pass out Chinese symbols
which are correct answers to the questions. The program enables the
person in the room to pass the Turing Test for understanding Chinese but
he does not understand a word of Chinese.
---
The point of the above argument is that, like a machine, the "chinese
room" is following explicit instructions, without understanding the
*meanings* of the symbols. This argument revolves around the notion that
we are semantical machines. That we understand MEANINGS in the way that
an algorithm cannot, since it is just following a mechanical procedure.
I would like to point out, that the true nature of our semantical
understanding is springing from completely *mechanical* (algorithmic)
process of learning. I suspect John Searle would appreciate my
counter-argument:
Semantical understanding is due to mechanical learning:
During our life we continuously learn things. We build "concepts" and
learn languages. When we learn new concepts, we base them on the
pre-existing concepts in our "worldview". When we learn new words, we
either look up their meaning from a dictionary through other words, or
understand the meaning of the new word through pre-learned concepts.
The matter of the fact is that in our conscious mind:
Words only have meaning relative to other words.
Concepts only have meaning relative to other concepts.
And this is so, because our learned worldview is not rooted to any
explicit knowledge of the world. It is at heart, a self-supporting
circle of beliefs:
When you were born, you obviously didn't know anything about anything.
You didn't know any words, you didn't understand any concepts. After
being brought into the world, your brain, a self-organizing learning
system, started a rigorous but mechanical process of learning, It tries
to seek some sense to all the strange sensory input.
Whatever assumptions it placed into the worldview that seemed to give
meaning to other assumptions, and vice versa, will become the
self-supporting root of your worldview. Don't underestimate the learning
ability of a fresh brain. It's a friggin' sponge I tell you! In fact,
that "you exist" is also just a semantical assumption you made at some
point of your life. It was only after that assumption that you could
store any memories that you interpetated as something occuring to you.
(That is why there is infant amnesia, and that is why you didn't really
even exist until your brain made assumptions as to what is "existence",
and that "I probably also do exist")
So while our learning process is completely mechanical, we are the
product of our experiences, and we don't truly understand anything in an
explicit sense. We are forced to see everything through semantics. Even
math is semantical to us. 6 x 7 = 42 just because we agreed on the rules
of how it is so. Counting such a thing in our head is not algorithmic
process like it is for a computer (if it was, you would not calculate
anything consciously).
We build a world-view this way, because it allows us to understand how
things work. This is same as understanding the meaning of things. When
we understand how things work, we can predict our environment, and
survive with our wits.
Because of this, we are forced to *interpetate the meaning* of
EVERYTHING we see around us.
Human communication can never be understood; it must always be interpetated.
Computer communication can never be interpetated; it must always be
interpetated.
But since this learning process IS explicit, then if we build a machine
that learns the same way, through explicit algorithm, it will not
consciously understand this learning process either, but IT WILL
UNDERSTAND SEMANTICS.
The problem with the chinese room argument is simply that it assumes no
learning takes place. The instruction table must not be given into the
room. The room must have ability to learn, and it must have "sensory
systems" that will give it information about the world that it is
supposed to try to make sense of. Then it will become to understand chinese.
It must be noted that while the worldview is a product of our
experiences, its interpetation is also a mechanical process.
Intentionality then, is a kind of illusion. Although you interpetate the
world in a way that you are consciously making decisions, and have free
will, the matter of the fact is that *you can only think of what you can
think of*, and your actions are in fact mechanically restricted by your
worldview, at all moments. You cannot base your decisions on anything
you haven't yet learned.
Anyone aspiring to build a truly conscious AI, should not put is
attention into how to program behaviour directly, but into how to make
the system learn in an open-ended fashion, without any true
understanding of anything.
If anyone knows Searle's e-mail address, I would be very interested to
hear what he might think about the above.
-Anssi
Anssi Hyytiainen
> Human communication can never be understood; it must always be
> interpetated.
> Computer communication can never be interpetated; it must always be
> interpetated.
Correcting a typo;
Computer communication can never be interpetated; it is always
understood exactly.
-Anssi
exama...@gmail.com
:)
Curt Welch
> Chinese Room argument goes as follows:
> ---
> Imagine a native English speaker who knows no Chinese locked in a room
> full of boxes of Chinese symbols together with a book of instructions
> for manipulating the symbols. Imagine that people outside the room send
> in other Chinese symbols which, unknown to the person in the room, are
> questions in Chinese. And imagine that by following the instructions in
> the program the man in the room is able to pass out Chinese symbols
> which are correct answers to the questions. The program enables the
> person in the room to pass the Turing Test for understanding Chinese but
> he does not understand a word of Chinese.
> ---
>
> The point of the above argument is that, like a machine, the "chinese
> room" is following explicit instructions, without understanding the
> *meanings* of the symbols. This argument revolves around the notion that
> we are semantical machines. That we understand MEANINGS in the way that
> an algorithm cannot, since it is just following a mechanical procedure.
>
> I would like to point out, that the true nature of our semantical
> understanding is springing from completely *mechanical* (algorithmic)
> process of learning. I suspect John Searle would appreciate my
> counter-argument:
>
> Semantical understanding is due to mechanical learning:
> If anyone knows Searle's e-mail address, I would be very interested to
> hear what he might think about the above.
I'm sure he would not. I'm sure he's heard far more than he's every wanted
to hear about the topic. There are many good websites that analyze the
different arguments related to the Chinese room. Here are two:
http://en.wikipedia.org/wiki/Chinese_room
http://plato.stanford.edu/entries/chinese-room/
The Chinese Room is not really an argument or a proof of anything. It's
best seen as an ink blot test designed to expose your own personal beliefs
about the mind. To attempt to argue it with someone is as futile as
attempting to argue whether an ink blot it a butterfly or a vagina.
People generally fall on one side of the debate or another, and nothing you
can say will cause someone to switch sides. It all comes down to what you
choose to believe "understand" means. Some of us, believe that a computer
program has the power to "understand" and some of us do not. If you
believe that a computer, when correctly programmed, will be able to
understand in the same way a human understands, then the Chinese Room
argument just looks stupid. If you believe a computer program could never
understand like a human understands, then the Chinese Room argument just
makes it clear why that's true.
The bottom line is that no one yet knows what understanding is even though
lots of people, including myself, like to believe we do. No one has yet
explained how the brain allows us to understand to the level needed for us
to duplicate that power in a machine. Without any of this hard evidence,
everything else is just speculation. The Chinese room doesn't change the
fact that we are all have our own personal opinions about what
understanding is.
Only hard evidence that shows what understanding really is will give anyone
the power to debate the truth of the Chinese room. Until that hard
evidence shows up, everyone will continue to pick the side they like best
and stick to it.
--
Curt Welch http://CurtWelch.Com/
cu...@kcwc.com http://NewsReader.Com/
Lucky
>room" is following explicit instructions, without >understanding the
>*meanings* of the symbols.
Define "Meaning".
>This argument
>revolves around the notion that
>we are semantical machines. That we >understand MEANINGS in the way that
>an algorithm cannot, since it is just following a >mechanical procedure.
>I would like to point out, that the true nature of >our semantical
>understanding is springing from completely >*mechanical* (algorithmic)
>process of learning. I suspect John Searle >would appreciate my
>counter-argument:
>Semantical understanding is due to >mechanical learning:
>During our life we continuously learn things. >We build "concepts" and
>learn languages. When we learn new >concepts, we base them on the
>pre-existing concepts in our "worldview".
Sounds like my theory of "Cognitive Paradigms as Foundation of Thought
and Comprehension"
Sorry, but I beat you to it.
You used "Pre-existing concepts" "worldview", etc.. These are examples
of paradigms.
I'll take it a step further....
The brain is incapable of comprehending new information without
connecting/super-imposing new information to a prior existing paradigm.
Using the example of an amoeba would suffice.
An amoeba extens a psuedopod to engulf and absorb food matter.
Comparing this to my theory, the brain "extends" a cognitive paradigm
in a similar fashion.
>When we learn new words, we
>either look up their meaning from a dictionary >through other words, or
>understand the meaning of the new word >through pre-learned concepts.
Incorrect. We learn new vocabulary through reading and subsequently
deducing meaning from context. I thought everyone knew this? W also
learn through repetition, neural reinforcement, but most children
identify negatively with such so the information is rarely stored long.
>The matter of the fact is that in our conscious >mind:
You are assuming it exists. My theory states it is an illusion.
>Words only have meaning relative to other >words.
Incorrect. Both words and that which they represent are firmly
connected to each other in the brain.
Ex.- Say "Purple" over and over outloud.
Eventualy you will cause the connection between the word "Purple" and
the color it symbolizes to disconnect. You will experience the word
"Purple" to be extremely silly and odd sounding. This is because you
no longer have the ability to comprehend a word once it is severed from
that from which it symbolizes. The good news is that as soon as you
start thinking about somehting else, the connections will reestablish
themselves. Again, this ability to "reconnect" is a part of my theory.
>Concepts only have meaning relative to other >concepts.
Incorrect. A "concept" is a paradigm. The human brain is incapable of
comprehending new information without applying a prior cognitive
paradigm to it. For example:
You are the first person to discover another type of bee inside a
beehive. You cannot comprehend the reasons for such nor can you
comprehend the role it plays. You start to gather information. You
make an observation that it is the only female in the hive that lays
eggs, and is thus fertile. The other bees pay extreme attention on
this bee. You unconsciously seek out prior paradigms to "capture" this
information and absorb it, turning it into comprehension. The correct
paradigm is selected and you yell out "Queen Bee, that's what I'll call
it!" Your brain reached out with the paradigm for "Queen" and
connected it to the role of the only fertile bee in the hive. Thus you
now have some level of comprehension. To be a bit more technical,
"Queen Bee" branches off to become it's own paradigm. For example, how
many times have we used this to refer to a hard working woman in leader
position?
Please do more research on paradigms. You will be delighted with
what you discover.
>And this is so, because our learned worldview >is not rooted to any
>explicit knowledge of the world. It is at heart, >a self-supporting
>circle of beliefs:
According to my theory, comprehension is indeed "rooted" in the
exterior world. Some aspects may be self-supporting, but you are
failing to take into consideration something you have do each and every
day unconsciously.
>When you were born, you obviously didn't >know anything about anything.
>You didn't know any words, you didn't >understand any concepts. After
>being brought into the world, your brain, a self->organizing learning
>system, started a rigorous but mechanical >process of learning, It tries
>to seek some sense to all the strange >sensory input.
I would not word it as such. As a newborn, there is no concern for
such. As the baby becomes a todler, then problems arise. The brain
does not "try to make sense" of sensory input. It flat out believes
it...that is until something causes it not to trust sensory input
alone. Again, this is covered by my theory.
>Whatever assumptions it placed into the >worldview that seemed to give
>meaning to other assumptions, and vice >versa, will become the
>self-supporting root of your worldview.
"Assumptions" is yet another example of a Paradigm.
Otherwise, some people can becoem "stuck" in a particuliar paradigm and
not be able to percieve anything differently. By the way, that's one of
the function paradigms, they shape our perceptions. I didn't originate
that theory though.
> Don't >underestimate the learning
>ability of a fresh brain.
Define "fresh". 2 weeks past date printed on cortex?
Don't underestimate the power fo the brain period.
> It's a friggin' sponge I >tell you!
Good Ole' "Plasticity".
>In fact,
>that "you exist" is also just a semantical >assumption you made at some
>point of your life.
Yes, as I replied above, my theory suggests that "consciousness" is
simply an illusion, compatible to a movie projector. Move the frames
fast and the illusion of reality and existence is produced.
> It was only after that >assumption that you could
>store any memories that you interpetated as >something occuring to you.
Incorrect. The brain has the ability to store 2 different types of
information.
1.) Information it understands.
2.) Information it doesn't understand.
The storage of information the brain doens't understand doesn't rely on
a prior "assumption" to take place.
>(That is why there is infant amnesia, and that >is why you didn't really
>even exist until your brain made assumptions >as to what is "existence",
>and that "I probably also do exist")
Very close to understanding something important. Won't tell you what
it is though cause I'm saving that for my paper.
>So while our learning process is completely >mechanical, we are the
>product of our experiences, and we don't truly >understand anything in an
>explicit sense. We are forced to see >everything through semantics. Even
>math is semantical to us. 6 x 7 = 42 just >because we agreed on the rules
>of how it is so.
Yes! Yes! Yes! I've tried for years to get people to realize that
science itself is merely a paradigm in which to percieve reality! The
problem is that if you include such statement in any paper you will be
immediately drawn and quartered.
>Counting such a thing in our >head is not >algorithmic
>process like it is for a computer (if it was, you >would not calculate
>anything consciously).
>We build a world-view this way, because it >allows us to understand how
>things work.
Incorrect. "world-view" is merely a paradigm. Paradigms shape
perception. My theory explains how Comprehension is created.
>This is same as understanding >the meaning >of things. When
>we understand how things work, we can >predict our environment, and
>survive with our wits.
>Because of this, we are forced to *interpetate >the meaning* of
>EVERYTHING we see around us.
Why would do that?
>Human communication can never be >understood; it must always be interpetated.
Incorrect. You are attempting to discover "context" but unfortunately
someone already beat you to it.
>Computer communication can never be >interpetated; it must always be
>interpetated.
?
>But since this learning process IS explicit, >then if we build a machine
>that learns the same way, through explicit >algorithm, it will not
>consciously understand this learning process >either, but IT WILL
>UNDERSTAND SEMANTICS.
Yes! Just like us. Stick that in your hat Psychologists!
>The problem with the chinese room argument >is simply that it assumes no
>learning takes place. The instruction table >must not be given into the
>room. The room must have ability to learn, >and it must have "sensory
>systems" that will give it information about the >world that it is
>supposed to try to make sense of. Then it will >become to understand chinese.
I Hate Turing.
>It must be noted that while the worldview is a >product of our
>experiences, its interpetation is also a >mechanical process.
Again, my theory.
>Intentionality then, is a kind of illusion. >Although you interpetate the
>world in a way that you are consciously >making decisions, and have free
>will, the matter of the fact is that *you can >only think of what you can
>think of*,
Wrong! I can't show you how to get around this becasue I am saving it
for my paper. Let me give you a hint. You wrote that because you were
using the wrong paradigm. You need to find the right paradigm that will
allow you to comprehend the way to "think beyond thought".
>and your actions are in fact >mechanically >restricted by your
>worldview, at all moments. You cannot base >your decisions on anything
>you haven't yet learned.
FYI, "World-Views", are paradigms that exist in a state of flux. Very
few of them are so extremely rigid. The problem is that they are
"invisible". We have to put clothes on them to see them. The clothes
are language/words.
>Anyone aspiring to build a truly conscious AI, >should not put is
>attention into how to program behaviour >directly, but into how to make
>the system learn in an open-ended fashion, >without any true
>understanding of anything.
Well, you sounded good until the very last bit. "without any true
understanding of anything"? If you do that then you are simply
creating another version of a Turing machine.
>If anyone knows Searle's e-mail address, I >would be very interested to
>hear what he might think about the above.
I would suggest clearing this up a whole lot before trying to submit it
further.
One of the problems you will encounter is because you are using
"paradigms" in an ignorant fashion, without knowledge of. You will
need to do more research on paradigms before proceding. Also keep in
mind that most scientists are going to reject any theory containing
paradigms simply because they know for a fact that if they accept such
a theory then it would mean their whole career was a waste.
-Jason
Lucky
>best seen as an ink blot test designed to expose your own personal beliefs
>about the mind. To attempt to argue it with someone is as futile as
>attempting to argue whether an ink blot it a butterfly or a vagina.
"The Chinese Room as Ink Blot" is simply a paradigm, a way to percieve
Turing's arguement. However, that doesn't limit the arguement to such.
The key is to point out any faulty logic Turing used.
>People generally fall on one side of the debate or another, and nothing you
>can say will cause someone to switch sides. It all comes down to what you
>choose to believe "understand" means. Some of us, believe that a computer
>program has the power to "understand" and some of us do not. If you
>believe that a computer, when correctly programmed, will be able to
>understand in the same way a human understands, then the Chinese Room
>argument just looks stupid. If you believe a computer program could never
>understand like a human understands, then the Chinese Room argument just
>makes it clear why that's true.
Well, I believe that it is possible for machines to become intelligent.
However, I don't believe Turing theory can be used to ceate
intelligence. I also don't like the direction the field is going.
Yet, it doesn't matter in the slightest because I am merely one person.
>The bottom line is that no one yet knows what understanding is even though
>lots of people, including myself, like to believe we do.
I do.;)
> No one has yet
>explained how the brain allows us to understand to the level needed for us
>to duplicate that power in a machine.
I do ;)
> Without any of this hard evidence,
>everything else is just speculation. The Chinese room doesn't change the
>fact that we are all have our own personal opinions about what
>understanding is.
I don't have any personal opinions about what "understanding" is. I
prefer to use "comprehension". My theory was based on a flash of
insight. I expounded
up on it and proceded from there. The thing is, becasue my theory is
so simple, I thought surely someone must have thought of it before me.
So I asked around and nope, nobody had any clue.
I have presented some of my thoughts here on another thread. However,
I am saving "The Whopper" for publication.
What suprises me is that I have already applied my theory to other
questions and have recieve very interesting results.
For example, "Why do I percieve my mind to be centered in my head?"
The standard explanations are that the brain has some self-awareness or
that the mind has a strong connection to the eyes.
Applying my theory :
"The mind is percieved to be centered within our head simply because
the head is the one part of the body in which the five senses converge.
If the senses are viewed as "windows" to the outside environment then
the head would be the part of the body with the best "view".
Accordingly, it wouldn't matter where the brain was placed inside the
body. If it was placed within the left big toe, for instance, the mind
would still be percieved to be centered within the head."
or how about this:
"Why is the average age of an American girls first menses decreasing?"
As far as I know, there is no standard explanation.
Applying my theory:
"The average age for an American girl's first menses is decreasing
because the number of overweight/obese American youth is rising. In
order for a girl to experience a menstrual cycle, she must currently
possess a pre-set amount of stored fat. If the girl is of average
weight or below weight, then even if the hormones and developement
allow a menses, it will not occur until the time that sufficient levels
of fat are indeed stored. If a girl is overweight/obese at the begining
of purberty then no such waiting period is necessary and the girl will
experience menses as soon as the hormones and developement allow."
>Only hard evidence that shows what understanding really is will give anyone
>the power to debate the truth of the Chinese room. Until that hard
>evidence shows up, everyone will continue to pick the side they like best
>and stick to it.
Come talk to me. I know. ;)
Hint: Psychologists have produced several theories of mind. All are
flawed in one single area. Figure out why they are flawed and then
follow the line of questioning from there. You will be led down a path
leading to a "Brave New World". If anybody ever gets there, drop me an
email. I'm lonely being the only one here so far.
-Jason
--
>Curt Welch
Anssi Hyytiainen
>>If anyone knows Searle's e-mail address, I would be very interested to
>>hear what he might think about the above.
>
> I'm sure he would not. I'm sure he's heard far more than he's every wanted
> to hear about the topic. There are many good websites that analyze the
> different arguments related to the Chinese room. Here are two:
>
> http://en.wikipedia.org/wiki/Chinese_room
>
> http://plato.stanford.edu/entries/chinese-room/
Hi Curt! :)
All the counter arguments I've heard of, including whatever is present
in the above websites, are just, well, dumb. I mean stuff like:
"Although the individual in the Chinese room does not understand
Chinese, perhaps the person and the room considered together as a system
do."
Um no.
Counter-arguments like that only prove that Searle's chinese room was
well placed to reveal a problem in what we think "knowledge" or
"intelligence" is. Look-up algorithm is not conscious no matter how you
do it.
The point of the counter argument is not to say "well perhaps if the
data is looked from the databank this or that way, then it's conscious".
The point is that we are NOT just a database from which we just look for
explicit answers. We are learning machines. *We don't have any answers!*
It doesn't attack the chinese room argument so much as it attacks the
idea it is based on; that semantics could be somehow intrinsic to our
brain "like metabolism etc". Such an idea can only arise from an
inability to see what else semantics could be.
Well, since we do not actually know anything for certain, instead we
only hold assumptions in our worldview, what else could arise but
semantics? Everything connects in our worldview. And by connection I
don't mean a low-level network connection, I mean higher abstraction, as
to what meaning one thing has to another thing.
(Obviously this stuff about semantics doesn't directly concern the
low-level self-organization, such as your networks)
I haven't come across similar counter-argument before. And it's pretty
sad...
> It all comes down to what you
> choose to believe "understand" means. Some of us, believe that a computer
> program has the power to "understand" and some of us do not.
Sure, you just have to make the distinction between explicit
understanding, such as what occurs in the brain at a low-level, or to
your networks, and semantical understanding, such as the potential
high-level behaviour of your network when trying to make sense of the
world tby some method. That method for us is simply learning.
> Only hard evidence that shows what understanding really is will give anyone
> the power to debate the truth of the Chinese room. Until that hard
> evidence shows up, everyone will continue to pick the side they like best
> and stick to it.
I think chinese room has its place in that, there are a lot of people
working on AI systems who clearly don't have a clue. Being able to come
over the chinese room argument; to understand what semantics have to do
with intelligence, is very important. (ok, it may not be for your
approach, but in principle it helps to understand this)
-Anssi
Anssi Hyytiainen
>>The point of the above argument is that, like a >machine, the "chinese
>>room" is following explicit instructions, without >understanding the
>>*meanings* of the symbols.
>
> Define "Meaning".
Meaning as in conscious understanding. Himan understanding of meaning of
language, or gestures, or concepts. Anything you can think of
consciously, is something you attach a meaning to. And the odds are,
that the meaning you attach to something, is slightly different from the
meaning someone else attaches to same thing.
>>During our life we continuously learn things. >We build "concepts" and
>>learn languages. When we learn new >concepts, we base them on the
>>pre-existing concepts in our "worldview".
>
> Sounds like my theory of "Cognitive Paradigms as Foundation of Thought
> and Comprehension"
> Sorry, but I beat you to it.
That's cool. It's not like I just realized this stuff. There are older
texts I've written about it. It's just that I just become so frustrated
in the inability of people to form a meaningful counter argument. I was
looking for one, but couldn't find any. It was just that same
meaningless blabbering about "well maybe this or that is conscious"
without any attention paid to how we cannot actually "know" anything the
way a computer algorithm knows.
The matter of the fact is, that the so-called Turing test, is horsesh**.
There are a wide array of possible systems that could pass it without
being conscious. Searle is right about that. Passing Turing test is
meaningless.
I would suggest another method of finding out if a machine is truly
intelligent. When your system learns without having any initial
knowledge of the world upon which it could base its learning, and still
it will eventually ask you where will it go when it dies. Then you'll know.
> You used "Pre-existing concepts" "worldview", etc.. These are examples
> of paradigms.
>
> I'll take it a step further....
>
> The brain is incapable of comprehending new information without
> connecting/super-imposing new information to a prior existing paradigm.
Yeah, exactly so.
>>either look up their meaning from a dictionary >through other words, or
>>understand the meaning of the new word >through pre-learned concepts.
>
> Incorrect. We learn new vocabulary through reading and subsequently
> deducing meaning from context.
Yes yes, that is exactly the idea I was trying to convey by saying
"understanding the meaning of the new word through pre-learned concepts".
That is the same thing as saying deducing the meaning from context.
Understanding the context here is where "pre-learned concepts" step into
the picture. In other words, I was simply saying that new
concepts/meanings are built on top of old concepts/meanings.
>>The matter of the fact is that in our conscious >mind:
> You are assuming it exists. My theory states it is an illusion.
Mine does too. I'm restricted by vocabulary and existing concepts.
Consciousness exists only as a process that can be said to be conscious.
There's nothing fundamental about it. Nothing that is "me", apart from
my memories. I think our views are pretty much identical. The
differences are probably just semantical, and pretty much prove the
point that human communication can never be understood, it must be
interpetated.
>>Words only have meaning relative to other >words.
> Incorrect. Both words and that which they represent are firmly
> connected to each other in the brain.
The point is just that you cannot explain what some word means, without
explaining this with other words. Words and concepts, they are really
just the same thing. The connection between word "purple" and a colour
purple is not fundamental, just like the connection between running and
moving the muscles of your feet is not fundamental. It is learned, even
though repetition places concepts under your conscious level. It does
this only when you have mastered the pattern so that you don't need to
"look for answers" from your worldview anymore.
>>Concepts only have meaning relative to other >concepts.
>
> Incorrect. A "concept" is a paradigm. The human brain is incapable of
> comprehending new information without applying a prior cognitive
> paradigm to it. For example:
>
> You are the first person to discover another type of bee inside a
> beehive. You cannot comprehend the reasons for such nor can you
> comprehend the role it plays. You start to gather information. You
> make an observation that it is the only female in the hive that lays
> eggs, and is thus fertile. The other bees pay extreme attention on
> this bee. You unconsciously seek out prior paradigms to "capture" this
> information and absorb it, turning it into comprehension. The correct
> paradigm is selected and you yell out "Queen Bee, that's what I'll call
> it!" Your brain reached out with the paradigm for "Queen" and
> connected it to the role of the only fertile bee in the hive. Thus you
> now have some level of comprehension. To be a bit more technical,
> "Queen Bee" branches off to become it's own paradigm. For example, how
> many times have we used this to refer to a hard working woman in leader
> position?
I think you are describing just what I'm thinking. You describe how I
might call something a queen bee. Of course the idea of a queen bee has
no meaning without other concepts, such as bees and queens. And they
don't have meaning without other concepts, etc, just like you described.
So what exactly is incorrect in stating "concepts only have meaning
relative to other concepts"? You probably just tripped onto a completely
semantical issue. Please don't :)
> Please do more research on paradigms. You will be delighted with
> what you discover.
Sounds like I'll discover it is just another semantical way to see the
same thing that I think springs consciousness :)
> According to my theory, comprehension is indeed "rooted" in the
> exterior world. Some aspects may be self-supporting, but you are
> failing to take into consideration something you have do each and every
> day unconsciously.
Anything I do unconsciously, is not something that is part of my
conscious experience.
Instincts play a part in triggering the learning processes (such as
curiosity). But they are not part of my conscious world view in that I
cannot use them as part of intelligent reasoning. (Apart from
understanding that they affect the behaviour of people, but that has
nothing to do with consciously understanding the meaning of "love" or
"fear")
> I would not word it as such. As a newborn, there is no concern for
> such. As the baby becomes a todler, then problems arise. The brain
> does not "try to make sense" of sensory input. It flat out believes
> it...
I think there are plenty of indications pointing otherwise. For example,
the way you see the whole world upright, is purely a matter of learning
that that's the better way to decipher the visual information. Like you
probably know, you can wear goggles that turn the world upside down in
your view, and if you wear them long enough, you have learned to "see"
the world upright again, without conscious effort.
And when you remove the goggles, hello upside-down world again.
So there's no "truth" as to which way around the world should be
"seen",other than which way seems to agree with other sensory input
(your balance organ, for example, as far as you understand what the
meaning of balance sensation is).
Other thing is, that there exist sensory implants, which simply
stimulate the appropriate parts of the brain according to a visual image
through a camera. Brain can learn to find the meaning from this
arbitrary input, and the person will experience a visual imagery. (Well,
very crude one with present technology, but still)
And if you think about it, it's pretty impossible to imagine how a baby
brain, that doesn't have any preconceptions of what is world like, what
existence is, what it means to see, hear, taste or smell things, or have
any knowledge of that sort, could understand what any sensory
stimulation it receives could possibly mean. It's all just some sort of
alien data, until you tag meaning to it. And soon you are deciphering
it. With no conscious effort.
>>Whatever assumptions it placed into the >worldview that seemed to give
>>meaning to other assumptions, and vice >versa, will become the
>>self-supporting root of your worldview.
>
> "Assumptions" is yet another example of a Paradigm.
I am kind of limited to use words. All words are paradigms/concepts.
> Otherwise, some people can becoem "stuck" in a particuliar paradigm and
> not be able to percieve anything differently. By the way, that's one of
> the function paradigms, they shape our perceptions. I didn't originate
> that theory though.
Is there more material about it somewhere?
>>Don't >underestimate the learning
>>ability of a fresh brain.
>
> Define "fresh". 2 weeks past date printed on cortex?
Heh.
Fresh brain is something that is starting to build its world view. Fresh
brain is empty. Doesn't know jack. The more the brain "knows", the less
it will be able to learn. It becomes full of scepticism, because most of
the information it comes across, seems false because of some
pre-existing piece of information.
Children on the other hand take in all kinds of nazi-propaganda without
hesitation.
> Yes, as I replied above, my theory suggests that "consciousness" is
> simply an illusion, compatible to a movie projector. Move the frames
> fast and the illusion of reality and existence is produced.
Cool. We agree.
>>It was only after that >assumption that you could
>>store any memories that you interpetated as >something occuring to you.
> Incorrect. The brain has the ability to store 2 different types of
> information.
>
> 1.) Information it understands.
> 2.) Information it doesn't understand.
>
> The storage of information the brain doens't understand doesn't rely on
> a prior "assumption" to take place.
In my vocabulary, "not understanding" is also learning. Any experience
is learning. Even non-experience is learning. If all your sensory
systems are cut, you will be in the dark wondering why you can't feel
anything. That is conscious experience, and you learn something. You
will have a memory of this experience of not experiencing anything.
Not being able to place a semantical meaning onto something could be
called "not understanding", but the experience of not understanding is
still conscious one.
And understanding is just a case of the person being able to attach the
experience into his existing world-view in such way, that he thinks he
has understood the subject matter. This thinking is also semantical. In
a strict sense, he hasn't understood it. He has only interpetated it in
a way that seems to make sense. He has placed "meaning" onto the experience.
>>(That is why there is infant amnesia, and that >is why you didn't really
>>even exist until your brain made assumptions >as to what is "existence",
>>and that "I probably also do exist")
>
> Very close to understanding something important. Won't tell you what
> it is though cause I'm saving that for my paper.
That one doesn't really ever exist, one just thinks he does? There's
nothing essentially us that stays with our body throughout our lives. If
you are to lose all your memories, a completely new person will come out
from the brain, if "persons" we may be called.
What paper? Perhaps I can beat you to that ;)
>>So while our learning process is completely >mechanical, we are the
>>product of our experiences, and we don't truly >understand anything in an
>>explicit sense. We are forced to see >everything through semantics. Even
>>math is semantical to us. 6 x 7 = 42 just >because we agreed on the rules
>>of how it is so.
>
> Yes! Yes! Yes! I've tried for years to get people to realize that
> science itself is merely a paradigm in which to percieve reality! The
> problem is that if you include such statement in any paper you will be
> immediately drawn and quartered.
Really? What sad numb-mindedness. I'll help you make them understand if
I can.
>>Because of this, we are forced to *interpetate >the meaning* of
>>EVERYTHING we see around us.
>
> Why would do that?
How can I not do that. Anything I think I "know", can be challenged
quite trivially by any philosopher. And then there's nothing I can do to
point them explicitly wrong.
>>Human communication can never be >understood; it must always be interpetated.
> Incorrect. You are attempting to discover "context" but unfortunately
> someone already beat you to it.
Hey you don't know when I started this ;)
>>Computer communication can never be >interpetated; it must always be
>>interpetated.
> ?
Unfortunate typo, sorry. I was just so frustrated in not finding any
meaningful counter arguments that I just typed something together too
quickly. Computer communication is explicit, that's what I was trying to
say. And because it is explicit, there's no room left for interpetation;
to meaning. 1 is 1, and 0 is 0, and that's that.
That is not to say, that an explicit, mechanical process couldn't
perform open-ended learning. It must be able to do this. And then it
will behave just like we do, and it will not really know anything. It
doesn't feel the bits that are flowing around inside it. But it will
contemplate the same questions that the greatest human thinkers have
contemplated. Such as if Batman is IN FACT better than Superman.
>>But since this learning process IS explicit, >then if we build a machine
>>that learns the same way, through explicit >algorithm, it will not
>>consciously understand this learning process >either, but IT WILL
>>UNDERSTAND SEMANTICS.
>
> Yes! Just like us. Stick that in your hat Psychologists!
Sounds like you've been frustrated by the same things that I have :) I
can't read any articles about congnition anymore without ripping the paper.
>>The problem with the chinese room argument >is simply that it assumes no
>>learning takes place. The instruction table >must not be given into the
>>room. The room must have ability to learn, >and it must have "sensory
>>systems" that will give it information about the >world that it is
>>supposed to try to make sense of. Then it will >become to understand chinese.
>
> I Hate Turing.
Hey, me too!
>>It must be noted that while the worldview is a >product of our
>>experiences, its interpetation is also a >mechanical process.
>
> Again, my theory.
Mine too! Quit hogging :)
>>Intentionality then, is a kind of illusion. >Although you interpetate the
>>world in a way that you are consciously >making decisions, and have free
>>will, the matter of the fact is that *you can >only think of what you can
>>think of*,
>
> Wrong! I can't show you how to get around this becasue I am saving it
> for my paper. Let me give you a hint. You wrote that because you were
> using the wrong paradigm. You need to find the right paradigm that will
> allow you to comprehend the way to "think beyond thought".
Oh, the mysterious paper. :P
So you think intentionality exists?
How do you suppose it works? Does it affect the physical state of the
brain somehow? What is intentionality anyway? Give me more hints you!
>>Anyone aspiring to build a truly conscious AI, >should not put is
>>attention into how to program behaviour >directly, but into how to make
>>the system learn in an open-ended fashion, >without any true
>>understanding of anything.
>
> Well, you sounded good until the very last bit. "without any true
> understanding of anything"? If you do that then you are simply
> creating another version of a Turing machine.
Nope. This is just a semantical issue, I'm sure you agree with the
point. I meant that the system must learn everything it knows in an
open-ended fashion. This means it will not hold any explicit knowledge
that couldn't be shaken by someone claiming this knowledge is wrong.
Turing machine is something that just follows an algorithm to produce
the behaviour, without any real learning taking place.
I am very much a prisoner of semantics. I may have used "true
understanding" in the meaning of semantical understaning, as in humans.
And sometimes I may have used it to mean the exact opposite "explicit
understanding". I hate semantics. I love semantics.
>>If anyone knows Searle's e-mail address, I >would be very interested to
>>hear what he might think about the above.
>
> I would suggest clearing this up a whole lot before trying to submit it
> further.
Heh, I think so too :) Like I said, I just got frustrated and wrote :)
> One of the problems you will encounter is because you are using
> "paradigms" in an ignorant fashion, without knowledge of. You will
> need to do more research on paradigms before proceding. Also keep in
> mind that most scientists are going to reject any theory containing
> paradigms simply because they know for a fact that if they accept such
> a theory then it would mean their whole career was a waste.
Maybe I'll call the paradigms concepts instead. They'll never see it
coming ;)
Thanks for the input! Gotta hit the sack. And by sack, I mean bed, and
I'm not actually going to hit it. Instead I'm going to sleep. I'm just
explaining this, in case you are a Turing machine :)
-Anssi
Lucky
Anssi Hyytiainen wrote:
> Lucky wrote:
> >>The point of the above argument is that, like a >machine, the "chinese
> >>room" is following explicit instructions, without >understanding the
> >>*meanings* of the symbols.
> >
> > Define "Meaning".
>
> Meaning as in conscious understanding. Himan understanding of meaning of
> language, or gestures, or concepts. Anything you can think of
> consciously, is something you attach a meaning to. And the odds are,
> that the meaning you attach to something, is slightly different from the
> meaning someone else attaches to same thing.
I accept meaning as having relavence to processes such as language,
where an abstract/symbol is used to represent an object or idea.
Otherwise, I shy from the word specificly due it's association of
having purpose, existence, or reality. I prefer to use the word
"comprehension" instead, since this communicates a more accurate
example. I may comprehend a subject but my comprehension in itself
does not validate the subject.
> >>During our life we continuously learn things. >We build "concepts" and
> >>learn languages. When we learn new >concepts, we base them on the
> >>pre-existing concepts in our "worldview".
> >
> > Sounds like my theory of "Cognitive Paradigms as Foundation of Thought
> > and Comprehension"
> > Sorry, but I beat you to it.
>
> That's cool. It's not like I just realized this stuff. There are older
> texts I've written about it. It's just that I just become so frustrated
> in the inability of people to form a meaningful counter argument. I was
> looking for one, but couldn't find any. It was just that same
> meaningless blabbering about "well maybe this or that is conscious"
> without any attention paid to how we cannot actually "know" anything the
> way a computer algorithm knows.
>
> The matter of the fact is, that the so-called Turing test, is horsesh**.
> There are a wide array of possible systems that could pass it without
> being conscious. Searle is right about that. Passing Turing test is
> meaningless.
I agree. I'm trying to formulate a test a person can administer if
they suspect they are dealing with a Turing Machine online. If I
succeed this will destroy any hope of "passing the A.I." test the
Turing's cultists may have. Currently I'm thinking along the lines of
a an "obvious answer" secret code, where a human will be able to pick
up the comunication with ease but a Turing Machine will not. One would
follow with asking the "person" if they could reply with the secret
word or phrase. If they can't ditch'em.
for example:
q w e q q a r e q q g o i n g q h o m e q
" We are going home"
Now, programers can always create a way around this, but humans will
always have the upper hand.
> I would suggest another method of finding out if a machine is truly
> intelligent. When your system learns without having any initial
> knowledge of the world upon which it could base its learning, and still
> it will eventually ask you where will it go when it dies. Then you'll know.
You would have to program "death".
People only question "death" when they encounter it through the loss of
a loved one, otherwise we gladly repress it.
There have been so many proposals for the test of intelligence. I'm
not sure what would work except for the ability to comprehend. I am
planing a system which mimics the cause of comprehension in humans,
thus giving rise to a true form of comprehension in A.I. It won't be
the same thing as human comprehension, but it will be a form of
comprehension none-the-less. I hate the concept of machines mimicking
intelligent replies and while not comprehending a single word they are
responding with.
Duh, I knew that. I wonder why it didn't dawn on me that is what you
were implying.
hmmm...try a different paradigm. We both know that we "see" things
upside-down, so we "assume" the brain turns it right-side up. It
could do that, however there are different solutions to the problem.
What if just the visual cortex itself is upside down? It would transmit
information the "correct' way in which we see.
> Like you
> probably know, you can wear goggles that turn the world upside down in
> your view, and if you wear them long enough, you have learned to "see"
> the world upright again, without conscious effort.
Yes, I've seen a science show with that experiment. However, I later
learned that it's not exactly what you think. You still "see" things
upside down. It's just the mind becomes use to it, so that "upside
down" becomes normal. It is your mind that turns upside down!
> And when you remove the goggles, hello upside-down world again.
NO! the whole googles thing takes like a few days to work. The effect
is not immediate. When you take them off for good your mind has to
adjust back to seeing right side up again!
My first introduction was with a cheesy self-help book along the lines
of "Change Your Paradigms, Chang Your Life!". I started to read it
and...hey, there's something about a paradigm....
Most of the other information I have found was not dedicated solely to
paradigms so you will have to dig through other stuff. Try Cognitive
Psychology or else do searches on Cognitive Paradigms.
>
> >>Don't >underestimate the learning
> >>ability of a fresh brain.
> >
> > Define "fresh". 2 weeks past date printed on cortex?
>
> Heh.
> Fresh brain is something that is starting to build its world view. Fresh
> brain is empty. Doesn't know jack. The more the brain "knows", the less
> it will be able to learn. It becomes full of scepticism, because most of
> the information it comes across, seems false because of some
> pre-existing piece of information.
Semantics: I believe the brain is indeed filled with "knowledge". It
just lacks comprehension.
>
> Children on the other hand take in all kinds of nazi-propaganda without
> hesitation.
>
> > Yes, as I replied above, my theory suggests that "consciousness" is
> > simply an illusion, compatible to a movie projector. Move the frames
> > fast and the illusion of reality and existence is produced.
>
> Cool. We agree.
>
> >>It was only after that >assumption that you could
> >>store any memories that you interpetated as >something occuring to you.
> > Incorrect. The brain has the ability to store 2 different types of
> > information.
> >
> > 1.) Information it understands.
> > 2.) Information it doesn't understand.
> >
> > The storage of information the brain doens't understand doesn't rely on
> > a prior "assumption" to take place.
>
> In my vocabulary, "not understanding" is also learning. Any experience
> is learning. Even non-experience is learning. If all your sensory
> systems are cut, you will be in the dark wondering why you can't feel
> anything. That is conscious experience, and you learn something. You
> will have a memory of this experience of not experiencing anything.
I disagree. I feel we can only think about that which we have some
level of comprehension of or are in the process of forming
comprehension of. I believe the word "thought" is a synonym for both
"consciousness" and "comprehension".
>
> Not being able to place a semantical meaning onto something could be
> called "not understanding", but the experience of not understanding is
> still conscious one.
"not understanding" is simply the ability to store information without
comprehension of such, the same as machines.
Our brains can store two types of information;
that which we comprehend
that which we do not comprehend
I wonder if this is connected to short & long term memory?
>
> And understanding is just a case of the person being able to attach the
> experience into his existing world-view in such way, that he thinks he
> has understood the subject matter. This thinking is also semantical. In
> a strict sense, he hasn't understood it. He has only interpetated it in
> a way that seems to make sense. He has placed "meaning" onto the experience.
No. I can't tell you why here because it involves what I am currently
working on.
There is a flaw in your logic above. You are unknowingly applying a
system to the one thing you can never successfuly apply it to. You
will always get answers that make sense to you but they will still be
incorrect. The problem is that most people never realize you would be
wrong.
>
> >>(That is why there is infant amnesia, and that >is why you didn't really
> >>even exist until your brain made assumptions >as to what is "existence",
> >>and that "I probably also do exist")
> >
> > Very close to understanding something important. Won't tell you what
> > it is though cause I'm saving that for my paper.
>
> That one doesn't really ever exist, one just thinks he does? There's
> nothing essentially us that stays with our body throughout our lives. If
> you are to lose all your memories, a completely new person will come out
> from the brain, if "persons" we may be called.
> What paper? Perhaps I can beat you to that ;)
I have a theory of comprehension. (Like who doesn't?) Anyway, my theory
engulfs all prior theories of mind, as well as proposes a "Brave New
World" literaly just beyond our comprehension.
One thing though, sensory input is vital to my theory of comprehension
though.
>
> >>It must be noted that while the worldview is a >product of our
> >>experiences, its interpetation is also a >mechanical process.
> >
> > Again, my theory.
>
> Mine too! Quit hogging :)
>
> >>Intentionality then, is a kind of illusion. >Although you interpetate the
> >>world in a way that you are consciously >making decisions, and have free
> >>will, the matter of the fact is that *you can >only think of what you can
> >>think of*,
> >
> > Wrong! I can't show you how to get around this becasue I am saving it
> > for my paper. Let me give you a hint. You wrote that because you were
> > using the wrong paradigm. You need to find the right paradigm that will
> > allow you to comprehend the way to "think beyond thought".
>
> Oh, the mysterious paper. :P
> So you think intentionality exists?
> How do you suppose it works? Does it affect the physical state of the
> brain somehow? What is intentionality anyway? Give me more hints you!
I chose not to place emphasis upon the theory of "Intentionality".
I am not concerned about that debate in the least.
your quote: "You can only think of what you can think of".
Hint: Your defintion of "think" is incorrect. Think =
comprehension/consciousness/understanding
We have the potential of understanding almost everything. There is one
thing we will never understand, yet if we attempt to do so we will
discover a theory concerning it that we will understand. The problem
is that it is guarenteed wrong because we are making an unconscious
assumption. "The Deceptive Consciousness" No more hints.
I like the way you think. I believe you would use the phrase, "Fresh
Mind".
-Jason
Glen M. Sizemore
as well as the interpretations of complex human behavior offerred by radical
behaviorists.
"Anssi Hyytiainen" <ans...@nic.fi.ANTISPAM> wrote in message
news:T55If.6791$oD5....@reader1.news.jippii.net...
Anssi Hyytiainen
> I accept meaning as having relavence to processes such as language,
> where an abstract/symbol is used to represent an object or idea.
> Otherwise, I shy from the word specificly due it's association of
> having purpose, existence, or reality. I prefer to use the word
> "comprehension" instead, since this communicates a more accurate
> example. I may comprehend a subject but my comprehension in itself
> does not validate the subject.
Yes, comprehension might be a better word for it. People usually assume
semantics is just referring to language (which is a semantical issue
also), while comprehension perhaps communicates it better that this has
to do with everything you can "think of".
So it has to do with everything that has traditionally been so difficult
for the Turing cult to implement. Like just moving in an environment
(walking, jumping, climbing, riding a bike, ice-skating, ski'ing) or
recognizing objects (solid, water, smoke, slippery ice) or faces
(mother, friend, man, smokin' hot chick). Of course the answer to all
these is that the machine needs to LEARN how to do these tricks.
(There is quite interesting disorder found from some people, where they
simply don't know how to recognize faces. They say recognizing faces is
like recognizing rocks)
> I'm trying to formulate a test a person can administer if
> they suspect they are dealing with a Turing Machine online. If I
> succeed this will destroy any hope of "passing the A.I." test the
> Turing's cultists may have. Currently I'm thinking along the lines of
> a an "obvious answer" secret code, where a human will be able to pick
> up the comunication with ease but a Turing Machine will not. One would
> follow with asking the "person" if they could reply with the secret
> word or phrase. If they can't ditch'em.
Yeah, or just have anything that is testing whether the person can learn
and pick up the meaning of some words. Like throw in phrases that make
sense for any person because they can pretty much guess the meaning,
even if they have never heard the phrase before. Or throw in some wicked
jokes or something. Or nasty remarks that need to be understood as a
whole before they are insulting. "You are an inspiration to birth
control" etc...
> You would have to program "death".
> People only question "death" when they encounter it through the loss of
> a loved one, otherwise we gladly repress it.
Ok, well, when they ask where people go when they die.
>>Like you
>>probably know, you can wear goggles that turn the world upside down in
>>your view, and if you wear them long enough, you have learned to "see"
>>the world upright again, without conscious effort.
>
> Yes, I've seen a science show with that experiment. However, I later
> learned that it's not exactly what you think. You still "see" things
> upside down. It's just the mind becomes use to it, so that "upside
> down" becomes normal. It is your mind that turns upside down!
That's exactly what I think it is. The person is struggling to
interpetate the image in correct way (consciously), and eventually
becomes so good at this that this "difficult" task happens
automatically, without thinking. Like the task of walking, or
ice-skating, or talking.
>>And when you remove the goggles, hello upside-down world again.
>
> NO! the whole googles thing takes like a few days to work. The effect
> is not immediate. When you take them off for good your mind has to
> adjust back to seeing right side up again!
That's what I said :) When you remove the goggles, you need to learn to
see other way around again.
I wonder if similar effect would happen in our process of interpetating
two images (both eyes) as one. What if someones other eye was blocked as
a child. Then when he grew older, we might remove the block, and ask him
if he sees any differently. I would bet my ass, he would report seeing
two images for quite a while.
Also, because you are so used to turning the flipped image of your
retina into "correct" orientation, then even when you've wear the
goggles for long enough for the image to appear "automatically" normal
to your consciousness again, you will persist to see all kinds of
anomalies in how objects appear to move in your view. Like a chair might
look upside down and then normally again in the next instant.
This is clearly because there is a vast number of ways how you actually
interpetate the visual information so that it makes most sense (you
don't see the blind spot, and you don't see the micro-movement of the
eyes that are there just to prevent neural adaptation, etc.), and these
other "corrective interpetations" that you perform to the image, are
working with the assumption that you DO see a flipped image. If you
would wear the goggles from a baby to adulthood, and THEN remove the
goggles, you would probably see similar anomalies for quite a while
after basically seeing the world in correct orientation again.
All this is really similar to thought experiments of having the nervous
systems to the muscles of your feet and legs mixed up. Sure it would be
almost impossible to walk and move around for a while, but eventually
you wouldn't even notice.
>>In my vocabulary, "not understanding" is also learning. Any experience
>>is learning. Even non-experience is learning. If all your sensory
>>systems are cut, you will be in the dark wondering why you can't feel
>>anything. That is conscious experience, and you learn something. You
>>will have a memory of this experience of not experiencing anything.
>
> I disagree. I feel we can only think about that which we have some
> level of comprehension of or are in the process of forming
> comprehension of.
Yeah, we are probably thinking about the same thing. "I don't understand
this" is also a form of comprehension. You comprehend that you don't
understand.
Perhaps your meaning of "not understanding" is something little bit
different. A bit deeper kind of non-understanding. Something that just
doesn't connect to our conscious mind at all.
>>And understanding is just a case of the person being able to attach the
>>experience into his existing world-view in such way, that he thinks he
>>has understood the subject matter. This thinking is also semantical. In
>>a strict sense, he hasn't understood it. He has only interpetated it in
>>a way that seems to make sense. He has placed "meaning" onto the experience.
>
> No. I can't tell you why here because it involves what I am currently
> working on.
>
> There is a flaw in your logic above. You are unknowingly applying a
> system to the one thing you can never successfuly apply it to. You
> will always get answers that make sense to you but they will still be
> incorrect. The problem is that most people never realize you would be
> wrong.
I'm not sure what you mean myself either, but perhaps I should clarify
that the point was simply, that if you have already learned all kinds of
things of the world, and you have reached consciousness in this process,
then even if all your sensory input is cut, you will find yourself in
the dark. There is nothing coming in through the sensory input to
comprehend, but you still comprehend that you are not feeling, seeing,
hearing or smelling anything.
And when the sensory input is restored, it's not like you were missing
consciousness just because of missing sensory input. You will tell the
doctors you were missing all sensory input for a while.
>>>I Hate Turing.
>>Hey, me too!
>
> One thing though, sensory input is vital to my theory of comprehension
> though.
Mine too. It is vital because one cannot begin to learn anything without
sensory input. And the less input there is, the more difficult it is to
build a meaningful worldview. The story if Helen Keller is enlightening.
>>>>Intentionality then, is a kind of illusion. >Although you interpetate the
>>>>world in a way that you are consciously >making decisions, and have free
>>>>will, the matter of the fact is that *you can >only think of what you can
>>>>think of*,
>>>
>>>Wrong! I can't show you how to get around this becasue I am saving it
>>>for my paper. Let me give you a hint. You wrote that because you were
>>>using the wrong paradigm. You need to find the right paradigm that will
>>>allow you to comprehend the way to "think beyond thought".
>>
>>Oh, the mysterious paper. :P
>>So you think intentionality exists?
>>How do you suppose it works? Does it affect the physical state of the
>>brain somehow? What is intentionality anyway? Give me more hints you!
>
> I chose not to place emphasis upon the theory of "Intentionality".
> I am not concerned about that debate in the least.
Well, perhaps you should be. Because it's another thing that people will
hit you in the head with. Since the laws of physics are explicit, how
can your brain, which is simply obeying laws of physics, have any true
way to control anything? It's thoughts, your behaviour. How can there be
intentionality in your behaviour?
The answer is that you don't have any true control in the sense of pure
intentionality. The method of open-ended learning -> building a
worldview -> interpetating the world through the worldview, explains why
you feel like you are in control, and "making decisions consciously".
Intentionality is just your predictions of how external systems work,
and what you think you should do about them. Whatever you think you
should do about them, is limited to whatever you know of the world.
> I like the way you think. I believe you would use the phrase, "Fresh
> Mind".
I think we are pretty much on the same track. It's good to know there
are people out there who really "comprehend" what's wrong with Turing
cult and why at its heart turing machine is still valid right idea. And
what is right about Chinese Room, and why it still fails in its ultimate
conclusion.
I just started to read the book of Jeff Hawkins, called "On
Intelligence". I haven't progressed very far, but it seems like he
thinks just the way we do too. He is expressing frustration to "Turing
cult" (well not in those words), and is working on some sort of AI
system as well.
Incidentally, his interest into true AI was sparked by the same American
Scientific brain issue that sparked my interest (although mine was the
re-issue, but still)
And in fact, on the page that I am just on while writing this, he is
talking about Chinese Room argument. And surprise surprise, he is
stating basically what I stated;
"As for me, I think Searle had it right. When I thought through the
Chinese Room argument and when I thought about how computers worked, I
didn't see understanding happening anywhere. I was convinced we needed
to understand what "understanding" is, a way to define it that would
make it clear when a system was intelligent and when it wasn't. It's
behaviour doesn't tell us this."
-Anssi
Immortalist
"Anssi Hyytiainen" <ans...@nic.fi.ANTISPAM> wrote in message
news:T55If.6791$oD5....@reader1.news.jippii.net...
> ---
> Imagine a native English speaker who knows no Chinese locked in a room
> full of boxes of Chinese symbols together with a book of instructions for
> manipulating the symbols. Imagine that people outside the room send in
> other Chinese symbols which, unknown to the person in the room, are
> questions in Chinese. And imagine that by following the instructions in
> the program the man in the room is able to pass out Chinese symbols which
> are correct answers to the questions. The program enables the person in
> the room to pass the Turing Test for understanding Chinese but he does not
> understand a word of Chinese.
> ---
>
> The point of the above argument is that, like a machine, the "chinese
> room" is following explicit instructions, without understanding the
> *meanings* of the symbols. This argument revolves around the notion that
> we are semantical machines. That we understand MEANINGS in the way that an
> algorithm cannot, since it is just following a mechanical procedure.
>
> I would like to point out, that the true nature of our semantical
> understanding is springing from completely *mechanical* (algorithmic)
> process of learning. I suspect John Searle would appreciate my
> counter-argument:
>
> Semantical understanding is due to mechanical learning:
>
Is this mechanical learning a hardwired instinct or an instinct who's
strength can be adjusted by experience at critical stages? I ask because
your text makes you sound like a blank slater who believes it not possible
for complex neural structures to come ready to be accented by the
environment.
Is it nature or nurture that makes us who we are? The question itself is a
false dichotomy;
[someone presents a situation as having only two alternatives, where in fact
other alternatives exist or can exist]
There are copious examples from human and animal behavior, which present the
notion that our environment affects the way our genes express themselves.
The switches controlling our 30,000 or so genes not only form the structures
of our brains but do so in such a way as to cue off the outside environment
in a tidy feedback loop of body and behavior.
We have genetic "thermostats" that are turned up and down by environmental
factors.
The proof is in the pudding for such touchy subjects as monogamy,
aggression, and parenting, which we now understand have some genetic
controls.
Nevertheless, the more we understand both our genes and our instincts, the
less inevitable they seem.
Not only are nature and nurture not mutually exclusive, but genes are
designed to take their cue from nurture.
Genes are not unchanging little bits of DNA: their expression varies
throughout a person's life, often in response to environmental stimuli.
Babies are born with genes hard-wired for sight, but if they are also born
with cataracts, the genes turn themselves off and the child will never
acquire the ability to see properly.
On the other hand, stuttering used to be ascribed solely to environmental
factors. Then stuttering was found to be clearly linked to the Y chromosome,
and evidence for genetic miswiring of areas in the brain that manage
language was uncovered. But environment still plays a role: not everyone
with the genetic disposition will grow up to be a stutterer.
Nature Via Nurture: Genes, Experience,
and What Makes Us Human, by Matt Ridley
http://www.amazon.com/exec/obidos/tg/detail/-/0060006781/
The Agile Gene: How Nature
Turns on Nurture, by Matt Ridley
http://www.amazon.com/exec/obidos/tg/detail/-/006000679X/
------------------------------------
"Homosexual men have a smaller clump of cells (known as INAH3) in the
hypothalamus than heterosexual men. The structure is closer in size to that
of women. This suggests, but does not establish, that sexual orientation is
a module determined by early brain development. If so, then the tendency of
heterosexual women to be sexually attracted to men and the tendency of
heterosexual men to be attracted to women may be attributable to anatomical
sex differences in the brain that are present at birth."
--Nigel Barber - Science of Romance
http://www.amazon.com/exec/obidos/tg/detail/-/1573929700/
-----------------------------------
Nurture is reversible; nature is not. That is the reason responsible
intellectuals have spent a century preferring the cheerful meliorism of
environment to the bleak Calvinism of genes. But what if there were a planet
where it was the other way around? Suppose some scientist discovered a world
in which lived intelligent creatures whose nurture was something they could
do nothing about, whereas their genes were exquisitely sensitive to the
world in which they lived...
...you [may] live on precisely such a planet...To the extent that people are
products of nurture, in the narrowly parental sense of the word, they are
largely the products of early and irreversible events. To the extent that
they are the product of genes, they are expressing new effects right into
adulthood, and often those effects are at the mercy of the way they live...
...it has long been clear from the experiences of homosexual people that
human sexual preferences are not only difficult to change but also fixed
from a very early age. Nobody in science now believes that sexual
orientation is caused by events in adolescence. Adolescence merely develops
a negative that was exposed much earlier. To understand why most men are
attracted to women while some men are attracted to men you must go much
further back into childhood, perhaps even into the womb.
The 1990s saw a series of studies that revived the idea of homosexuality as
a "biological" rather than a psychological condition, as a destiny rather
than a choice. There were studies showing that future homosexuals had
different personalities in childhood, studies showing that homosexual men
had differences in brain anatomy from heterosexual men, several twin studies
showing that homosexuality was highly heritable in western society, and
anecdotal reports from homosexual men to the effect that they had felt
"different" early in life." On its own none of these studies was
overwhelming. But together, and set against decades of proof that aversion
therapy, "treatment," and prejudice entirely failed to "cure" people of gay
instincts, the studies were emphatically clear. Homosexuality is an early,
probably prenatal, and irreversible preference. Adolescence simply throws
fuel on the fire.
What exactly is homosexuality? It is plainly a whole range of behavioral
characteristics. In some ways gay men seem to be more like women: they are
attracted to men, they may pay more attention to clothes, they are often
more interested in people than, say, football. In other ways, however, they
are more like heterosexual men: they buy pornography and seek casual sex,
for example. (Playgirl's nude centerfolds of men turned out to appeal mainly
to gays, not the intended women.)
People, like all mammals, are naturally female unless masculinized. Female
is the "default sex" (it is the other way around in birds). A single gene,
called SRY, on the Y chromosome starts a cascade of events in the developing
fetus leading to the development of masculine appearance and behavior. If
that gene is absent, a female body results. It is therefore reasonable to
hypothesize that homosexuality in men results from the partial failure of
this prenatal masculinization process in the brain, though not in the body
(see chapter 9).
By far the most reliable discovery about the causes of homosexuality in
recent years is Ray Blanchard's theory of the fraternal birth order. In the
mid-1990s Blanchard measured the number of elder brothers and sisters of gay
men compared with the population average. He found that gay men are more
likely to have elder brothers (but not elder sisters) than either gay women
or heterosexual men. He has since confirmed this in 14 different samples
from many different places. For each extra older brother, a man's
probability of being gay rises by one-third. (This does not mean that men
with many elder brothers are bound to be gay: an increase from, say, 3
percent of the population to 4 percent is an increase of one-third.)
Blanchard calculates that at least one gay man in seven, probably more, can
attribute his sexual orientation to this effect of fraternal birth order. It
is not simply birth order, because having elder sisters has no such effect.
Something about elder brothers must actually be causing homosexuality in
men. Blanchard believes the mechanism is in the womb rather than the family.
One clue lies in the birth weight of baby boys who will later become
homosexual. Normally, a second baby is heavier than a first baby of the same
sex. Boys especially are heavier if they are born after one or more sisters.
But boys born after one brother are only slightly heavier than firstborn
boys, and boys born after two or more brothers are usually smaller than
first-and second-born boys at birth. By analyzing questionnaires given to
gay and straight men and their parents, Blanchard was able to show that
younger brothers who went on to become homosexual were 170 grams lighter at
birth than younger brothers who went on to become heterosexual. He confirmed
the same result-high birth order, low birth weight compared with controls-in
a sample of 250 boys (with an average age of seven) who were showing
sufficient "cross-gender" wishes to have been referred to psychiatrists;
cross-gender behavior in childhood is known to predict later homosexuality.
Like Barker, Blanchard believes that conditions in the womb are marking the
baby for life. In this case, he argues, something about occupying a womb
that has already held other boys occasionally results in reduced birth
weight, a larger placenta (presumably in compensation for the difficulty the
baby experiences in growing), and a greater probability of homosexuality.
That something, he suspects, is a maternal immune reaction. The immune
reaction of the mother, primed by the first male fetuses, grows stronger
with each male pregnancy. If it is mild, it causes only a slight reduction
in birth weight; if strong, it causes a marked reduction in birth weight and
an increased probability of homosexuality.
What could the mother be reacting to? There are several genes expressed only
in males, and some are already known to raise an immune reaction in mothers.
Some are expressed prenatally in the brain. One intriguing new possibility
is a gene called PCDH22, which is on the Y chromosome, is therefore specific
to males, and is probably involved in building the brain. It is the recipe
for a protocadherin (see Chapter 5). Could this be the gene that wires the
bit of the brain that is peculiar to males? A maternal immune reaction may
be sufficient to prevent the wiring of the part of the brain that would
eventually encourage a fascination with female bodies.
Clearly not all homosexuality is caused this way. Some of it may be caused
directly by genes in the homosexual person without the mediation of the
mother's immune reaction. Blanchard's theory may explain why it has proved
so hard to pin down the "gay gene." The main method for finding such a gene
is to compare markers on the chromosomes of homosexual men with those of
their heterosexual brothers. But if many gay men have straight elder
brothers, this method would work poorly. Besides, the key genetic difference
might be on the mother's chromosomes, where it causes the immune reaction.
This might explain why homosexuality looks as though it is inherited through
the female line: genes for a stronger maternal immune reaction could appear
to be "gay genes," even though they may not be expressed in the gay man
himself but only in the mother.
But notice what this does to nature versus nurture. If nurture, in this case
birth order, causes some homosexuality, it does so by causing an immune
reaction, which is a process directly mediated by genes. So is this that
environmental or genetic? It hardly matters, because the absurd distinction
between reversible nurture and inevitable nature has now been well and truly
buried. Nurture in this case looks just as irreversible as nature, perhaps
more so.
Politically, the confusion is even greater. Most homosexuals welcomed the
news in the mid-1990s that their sexual orientation looked "biological."
They wanted it to be a destiny, not a choice, because that would undermine
the argument of homophobes that it was a choice and therefore morally
questionable. How could it be wrong if it was innate? Their reaction is
understandable but dangerous. A greater tendency to violence is also innate
in the human male. That does not make it right. Reasoning that "ought" can
be derived from "is" is called "naturalistic fallacy." To base any moral
position on a natural fact, whether that fact is derived from nature or from
nurture, is asking for trouble. In my morality, and I hope in yours, some
things are bad but natural, like dishonesty and violence; others are good
but less natural, like generosity and fidelity.
- THROWING SWITCHES IN THE BRAIN
It is easy to infer the existence of critical periods during which the wet
cement of character can be set. It is less easy to conceive of how they
work...
...- YOUNG TONGUES
Critical-period imprinting is everywhere. There are a thousand ways in which
human beings are malleable in their youth, but fixed once adult. Just as a
gosling is imprinted with an image of its mother during the hours after
birth, so a child is imprinted with everything from the number of sweat
glands on its body and a preference for certain foods to an appreciation of
the rituals and patterns of its own culture. Neither the gosling's
mother-image nor the child's culture is in any sense innate. But the ability
to absorb each is.
An obvious example is accent. People change their accents easily during
youth, generally adopting the accent of people of their own age in the
surrounding society. But sometime between about 15 and 25, this flexibility
simply vanishes. From then on, even if a person emigrates to a different
country and lives there for many years, his or her accent will change very
little. People may pick up a few inflections and habits from their new
linguistic surroundings, but not many. This is true of regional as well as
national accents: adults retain the accent of their youth; youngsters adopt
the accent of the surrounding society...
Nature Via Nurture: Genes, Experience, and What Makes Us Human by Matt
Ridley http://www.amazon.com/exec/obidos/tg/detail/-/0060006781/
Lester Zick
<ans...@nic.fi.ANTISPAM> in comp.ai.philosophy wrote:
>Chinese Room argument goes as follows:
>---
>Imagine a native English speaker who knows no Chinese locked in a room
>full of boxes of Chinese symbols together with a book of instructions
>for manipulating the symbols. Imagine that people outside the room send
>in other Chinese symbols which, unknown to the person in the room, are
>questions in Chinese. And imagine that by following the instructions in
>the program the man in the room is able to pass out Chinese symbols
>which are correct answers to the questions. The program enables the
>person in the room to pass the Turing Test for understanding Chinese but
>he does not understand a word of Chinese.
It occurs to me there is an implicit assumption here which needs to be
clarified. In what language are the instructions written?
Obviously they can't be written in Chinese because the English speaker
doesn't understand Chinese. So saying the instructions are written in
English or any other language understood by the English speaker and
gives uniformly correct answers to questions in Chinese means Chinese
is exhaustively subsumed by some other language.
Is this instruction language a universal language? If not there is no
reason to suggest it can be used to interpret Chinese correctly. If so
there is no proof such a language is in fact universal. And so far as
I know Turing "language" only applies to computable numbers and
there is no proof of equivalence between computable "language" in
turing terms and human language in general.
I don't see any special merit to the rest of these observations. It
looks like you're trying to suggest symbol translation is naive and
simplistic but human language is variegated and complex. Symbol
translation can certainly be simplistic but there is nothing to
suggest it has to be especially when heuristic significance is taken
into account. I think the more interesting and definitive question is
whether human language is equivalent to computable numbers.
~v~~
Lucky
strength can be adjusted by experience at critical stages? I ask
because
your text makes you sound like a blank slater who believes it not
possible
for complex neural structures to come ready to be accented by the
environment.
Not Anssi, but will throw my 2 cents in...
****************************************************************************************
Is it nature or nurture that makes us who we are? The question itself
is a
false dichotomy;
[someone presents a situation as having only two alternatives, where in
fact
other alternatives exist or can exist]
************************************************************************************
I don't believe Anssi brought up this arguement so I feel it would be
mute to
discuss so now.
*********************************************************************************
There are copious examples from human and animal behavior, which
present the
notion that our environment affects the way our genes express
themselves.
************************************************************************************
Yes, but only during natural development, unless you are refering to
diseases like canser, diabities, etc.
****************************************************************************************
The switches controlling our 30,000 or so genes not only form the
structures
of our brains but do so in such a way as to cue off the outside
environment
in a tidy feedback loop of body and behavior.
We have genetic "thermostats" that are turned up and down by
environmental
factors.
The proof is in the pudding for such touchy subjects as monogamy,
aggression, and parenting, which we now understand have some genetic
controls.
Nevertheless, the more we understand both our genes and our instincts,
the
less inevitable they seem.
Not only are nature and nurture not mutually exclusive, but genes are
designed to take their cue from nurture.
Genes are not unchanging little bits of DNA: their expression varies
throughout a person's life, often in response to environmental stimuli.
*********************************************************************************
Incorrect. Anytime the genes act outside their prescribed function it
is considered a disease. Genese associated with normal development are
turned off and remain off after they have served their function. A
person's eyes, for example, may slightly change color over a lifespan
due to environmental factors, but the genes responsible for coloring
the iris will not suddenly switch on at a later time producing another
shade.
*******************************************************************************
Babies are born with genes hard-wired for sight, but if they are also
born
with cataracts, the genes turn themselves off and the child will never
acquire the ability to see properly.
************************************************************
Cataracts rarely produce complete blindness. The genes do not turn
themselves off as you claim above. The process is called "Neural
Reinforcement". The genes give rise to the neural pathway for vission.
However, there is a catch. The catch is there is only a "window of
opportunity" for such. If the time passes with the neural pathway
going unstimulated then the brain's "plasticity" will take over and
divert the pathway for another use. Neural Reinforcement means that
the more a neural pathway is stimulated, the more the connection is
reinforced, making it almost "permanent". Neural Reinforcement
explains why we are right or left handed, why we form habbits, and has
even been sugested to be responsible for our sexual preferences.
*********************************************************************************
On the other hand, stuttering used to be ascribed solely to
environmental
factors. Then stuttering was found to be clearly linked to the Y
chromosome,
and evidence for genetic miswiring of areas in the brain that manage
language was uncovered. But environment still plays a role: not
everyone
with the genetic disposition will grow up to be a stutterer.
************************************************************************************
You are thinking about it wrong. Change your paradigm. Instead of
linking
stuttering specificly to males, which is incorrect, form the paradigm
of "disorder"
one in which men are most succeptable to.
BTW, there are women sufferes of stuttering as well. Though most
sufferers are men.
********************************************************************************
Nature Via Nurture: Genes, Experience,
and What Makes Us Human, by Matt Ridley
http://www.amazon.com/exec/obidos/tg/detail/-/0060006781/
The Agile Gene: How Nature
Turns on Nurture, by Matt Ridley
http://www.amazon.com/exec/obidos/tg/detail/-/006000679X/
------------------------------------
"Homosexual men have a smaller clump of cells (known as INAH3) in the
hypothalamus than heterosexual men. The structure is closer in size to
that
of women. This suggests, but does not establish, that sexual
orientation is
a module determined by early brain development. If so, then the
tendency of
heterosexual women to be sexually attracted to men and the tendency of
heterosexual men to be attracted to women may be attributable to
anatomical
sex differences in the brain that are present at birth."
*****************************************************************************
I was previously of the opinion that this study was dismissed mostly
due to bias of the researchers who were mostly gay men.
Also the study has major flaws.
For one, it assumes all gay men are of the stereotypical effiminate and
"flaming" environment.
Also, the hypothalmus is connected to behavior and does play a role
with regard to sexual hormones. The problem is that there are no
connections made between the hypothalmus and sexual preferences to
begin with. Nowhere in the study is there information presented as to
how the Hypothalmus may determine sexual preference. Until this is
presented then the notion that sexual preference and sex itself must
remain seperated.
***************************************************************
-----------------------------------
********************************************
Again, sexual preference is "hard-wired" simply because of the process
of Neural Reinforcement. A "sexual preference blank slate" can be made
to display any preference potential, be it standard like hertero &
homo, or criminal such as child molestor, bestiality, incest etc...
If we follow the line of reason you present, it soon becomes obvious
that you would need multiple theories to explain sexual preference.
Why choose multiply when one suffices?
According to Neural Reinforcement, a child contains neural pathways for
"sex" and not specific to any preference. As the child develops, their
desires, especialy during masturbation, influence the neural pathway.
The concept of "Window of Time" is also heavily connected to Neural
Reinforcement. As long as the child remains inside this "window" then
the child still has a choice and chance of altering the pathways in
whatever direction they may choose, assuming they are conscious of this
choice. (Most people are not conscious of having the choice nor even
comprehending a choice in preference at that age.) After the window of
time has passed, the neural pathways "freeze" and become hard-wired.
>From this point on, the child cannot help what he/she is sexually
attracted to. Now what needs to be stressed is the fact that there is
nothing wrong with homosexuality as a choice. The problem is that we
let gay people themselves trick us into thinking it's not a choice.
Well according to logic, the individual would never be conscious of
this choice in the first place!
The theory also explains why child molesters, and other deviant
sexuality can't be "cured".
***************************************************************************************
The 1990s saw a series of studies that revived the idea of
homosexuality as
a "biological" rather than a psychological condition, as a destiny
rather
than a choice. There were studies showing that future homosexuals had
different personalities in childhood, studies showing that homosexual
men
had differences in brain anatomy from heterosexual men, several twin
studies
showing that homosexuality was highly heritable in western society, and
anecdotal reports from homosexual men to the effect that they had felt
"different" early in life." On its own none of these studies was
overwhelming. But together, and set against decades of proof that
aversion
therapy, "treatment," and prejudice entirely failed to "cure" people of
gay
instincts, the studies were emphatically clear. Homosexuality is an
early,
probably prenatal, and irreversible preference. Adolescence simply
throws
fuel on the fire.
*******************************************************************
I believe the "window of time" on sexual preference passes far before
puberty begins. If my take on it is right then it should occur right
around the time the child begins to form "self", since the language
used is so descriptive of this process;
"I knew that I was different at a very early age.", is a statement made
by many gay people.
*********************************************************************************
*****************************************************************************
What exactly is homosexuality? It is plainly a whole range of
behavioral
characteristics. In some ways gay men seem to be more like women: they
are
attracted to men, they may pay more attention to clothes, they are
often
more interested in people than, say, football. In other ways, however,
they
are more like heterosexual men: they buy pornography and seek casual
sex,
for example. (Playgirl's nude centerfolds of men turned out to appeal
mainly
to gays, not the intended women.)
******************************************************************************************
Why are you so fixated on "gay"? You are letting your mind be
controlled by paradigms which alter your perceptions. There are just
as many "masculine' gay men as there are effeminate ones. The problem
is that you can't tell the difference because in your mind you equate
"Flaming" as being equal to gay. Hell, in Highschool, our quarterback
was secretly gay!
I think you really need to check yourself befor typing anything similar
to the above ever again. There are way too many stereotypes in your
statements. I would refer you to my theory of "Cognitive Paradigms" as
the basis for thought and comprehension. I would aslo suggest you seek
out some gay friends so that you can actualy learn and experience that
they are just regular people.
Besides, do you realzie that up until modern times, there was no such
thing as "gay" in the first place! The ancient world held no labels
for sexual preference. In Greece for example, the deciding factor on
wether a man slept with a female prostititue or a male prostitute would
be who was "prettier" or more beautiful. That was the factor of
emphasis, not their genitalia or behavior. Now in the modern age, more
and more people are willingly accepting labels of preference as
definition of themselves.
**************************************************************************************
People, like all mammals, are naturally female unless masculinized.
Female
is the "default sex" (it is the other way around in birds). A single
gene,
called SRY, on the Y chromosome starts a cascade of events in the
developing
fetus leading to the development of masculine appearance and behavior.
If
that gene is absent, a female body results. It is therefore reasonable
to
hypothesize that homosexuality in men results from the partial failure
of
this prenatal masculinization process in the brain, though not in the
body
(see chapter 9).
******************************************************************************
OMG, you are completely ignorant on this subject. People are not
naturally female as you claim. What a ridiculous statement. Do a
google search on development of gender. You will realize that the
genitals for either a girl or boy develop at the exact same time of
gestations. Girl parts don't turn into boy parts, instead a "lump"
become differentiated into either vulva or penis/scrotum.
There was a study that claimed we all started out as being female. It
was quickly dismissed, yet again, for the bias of the researcher, a
woman whom it turned out was a quote, "Lesbian Feminist" and was
involved with various protesting groups whose agenda was completely
anti-male with regards to place in society. I hope this is not the
study you are refering to.
The genes that create both external and internal reproductive organs
are switched on at the same time. Also, the only thing the Y
chormosome does is basicly to increase the level of androgens, such as
testosterone. Guess what! The "lump" that diferentiates to produce
either female or male organs, get's it's instructions from the genes.
The genes responsible for both genders are found on the X
chromosone!!!!! I told you, the only role the Y chromosone does is to
increase the level of androgens. The reason why is tha the "lump" that
diferentiates into either female or male gentitals gets it's cue from
hormonal levels. If more estrogens are produced, XX, then the lump
develops into female organs. If more androgens are produced, XY, then
the lump develops into male organs! Sometimes, if the hormonal levels
are extremely similar, "inter-sexed" individuals are formed,
"Hermaphrodites", despite wether they are XX or XY! Yes, there are
some people walking this earth who were born with a penis but have XX
chromosones!!!!
**********************************************************************************************
By far the most reliable discovery about the causes of homosexuality in
recent years is Ray Blanchard's theory of the fraternal birth order. In
the
mid-1990s Blanchard measured the number of elder brothers and sisters
of gay
men compared with the population average. He found that gay men are
more
likely to have elder brothers (but not elder sisters) than either gay
women
or heterosexual men. He has since confirmed this in 14 different
samples
from many different places. For each extra older brother, a man's
probability of being gay rises by one-third. (This does not mean that
men
with many elder brothers are bound to be gay: an increase from, say, 3
percent of the population to 4 percent is an increase of one-third.)
**************************************************************************************
Yes, and phrenologists believe you can determine the personality of a
patient, even diagnose disease, by examining the bumbs on their head.
Both are completely ridiculous. Why are you so obsessed with "Gay"?
*********************************************************************************************
*********************************************************************************
All this is B.S. There is no part of the brain "peculiar" to males.
Women and Men have the exact same brain structures used for the exact
same reason. The only differences noted are gender diferences among
certain stuctures such as "eye and hand cordination" for men, and
"Language/communication" for women.
*******************************************************************************
Clearly not all homosexuality is caused this way. Some of it may be
caused
directly by genes in the homosexual person without the mediation of the
mother's immune reaction. Blanchard's theory may explain why it has
proved
so hard to pin down the "gay gene." The main method for finding such a
gene
is to compare markers on the chromosomes of homosexual men with those
of
their heterosexual brothers. But if many gay men have straight elder
brothers, this method would work poorly. Besides, the key genetic
difference
might be on the mother's chromosomes, where it causes the immune
reaction.
This might explain why homosexuality looks as though it is inherited
through
the female line: genes for a stronger maternal immune reaction could
appear
to be "gay genes," even though they may not be expressed in the gay man
himself but only in the mother.
******************************************
The "Gay" gene is B.S.
If you buy into this tripe, then you also have to assume there is a
gene for heterosexuality, pedophilia, necrophilia, bestiality, etc...
Again, a load of B.S.
************************************************************************
work...
...- YOUNG TONGUES
************************************************************************************8
I'm seriously questioning your motives in evern bringing up the topic
of
homosexuality. Since this subject is not connected to this thread, why
have
you done so?
I am hoping that you didn't have "trollish" desires of instigating a
fight.
-Jason
Immortalist
"Lucky" <Lucky...@aol.com> wrote in message
news:1139947075.1...@o13g2000cwo.googlegroups.com...
************************************************************************************
> I don't believe Anssi brought up this arguement so I feel it would be
> mute to discuss so now.
Please adjust your news reader because its to hard to see the difference
between who says what.
When Anssi said;
>> Semantical understanding is due to mechanical learning:
I wondered if he meant that it was instinctual or learned only or
conditioned instincts.
If you were typing anything else it is hard to tell in my newsreader so
please re-argue anything important that I have taken out because of your
newsreaders problems.
Lucky
>please re-argue anything important that I have taken out because of your
>newsreaders problems.
If you aren't able to distinguish between your own words and those of
another, then it woud be pointless for me to reply to your request in
the first place.
-Jason
Curt Welch
> The matter of the fact is, that the so-called Turing test, is horsesh**.
> There are a wide array of possible systems that could pass it without
> being conscious. Searle is right about that. Passing Turing test is
> meaningless.
>
> I would suggest another method of finding out if a machine is truly
> intelligent. When your system learns without having any initial
> knowledge of the world upon which it could base its learning, and still
> it will eventually ask you where will it go when it dies. Then you'll
> know.
In other words, you study it's behavior, and if it acts like a human, you
know it's intelligent? How is this test not identical to the Turing test?
Lucky
Anssi Hyytiainen wrote:
> Lucky wrote:
> > I accept meaning as having relavence to processes such as language,
> > where an abstract/symbol is used to represent an object or idea.
> > Otherwise, I shy from the word specificly due it's association of
> > having purpose, existence, or reality. I prefer to use the word
> > "comprehension" instead, since this communicates a more accurate
> > example. I may comprehend a subject but my comprehension in itself
> > does not validate the subject.
>
> Yes, comprehension might be a better word for it. People usually assume
> semantics is just referring to language (which is a semantical issue
> also), while comprehension perhaps communicates it better that this has
> to do with everything you can "think of".
>
> So it has to do with everything that has traditionally been so difficult
> for the Turing cult to implement. Like just moving in an environment
> (walking, jumping, climbing, riding a bike, ice-skating, ski'ing) or
> recognizing objects (solid, water, smoke, slippery ice) or faces
> (mother, friend, man, smokin' hot chick). Of course the answer to all
> these is that the machine needs to LEARN how to do these tricks.
Semantics again, replace "learn" with "comprehend". Without
comprehension
then then a machine nor person, can recognize a problem to begin with.
Remember my comment that we can only think of what we can comprehend,
otherwise
we just accept information at face value.
Interesting! I wonder if he might not be blind at first, only slowly
developing sight in that eye. It's also possible that the eye will
never be able to see.
>
> Also, because you are so used to turning the flipped image of your
> retina into "correct" orientation, then even when you've wear the
> goggles for long enough for the image to appear "automatically" normal
> to your consciousness again, you will persist to see all kinds of
> anomalies in how objects appear to move in your view. Like a chair might
> look upside down and then normally again in the next instant.
>
> This is clearly because there is a vast number of ways how you actually
> interpetate the visual information so that it makes most sense (you
> don't see the blind spot, and you don't see the micro-movement of the
> eyes that are there just to prevent neural adaptation, etc.), and these
> other "corrective interpetations" that you perform to the image, are
> working with the assumption that you DO see a flipped image. If you
> would wear the goggles from a baby to adulthood, and THEN remove the
> goggles, you would probably see similar anomalies for quite a while
> after basically seeing the world in correct orientation again.
>
> All this is really similar to thought experiments of having the nervous
> systems to the muscles of your feet and legs mixed up. Sure it would be
> almost impossible to walk and move around for a while, but eventually
> you wouldn't even notice.
>
> >>In my vocabulary, "not understanding" is also learning. Any experience
> >>is learning. Even non-experience is learning. If all your sensory
> >>systems are cut, you will be in the dark wondering why you can't feel
> >>anything. That is conscious experience, and you learn something. You
> >>will have a memory of this experience of not experiencing anything.
Semantics, what you label "not understanding", I would label "low level
of comprehension of". Back to my theory that we can only think of that
wich we
comprehend. Comprehension comes in all levels, and as long as our
paradigms are closed, then knew knowledge can constantly be added to
the string of thought.
> >
> > I disagree. I feel we can only think about that which we have some
> > level of comprehension of or are in the process of forming
> > comprehension of.
>
> Yeah, we are probably thinking about the same thing. "I don't understand
> this" is also a form of comprehension. You comprehend that you don't
> understand.
Yeah, again with "Low level of comprehension of".
>
> Perhaps your meaning of "not understanding" is something little bit
> different. A bit deeper kind of non-understanding. Something that just
> doesn't connect to our conscious mind at all.
Just what I said above, semantics. Better to say "low level of
comprehension of"
plus "I can't fit this well with the paradigm I am using".
>
> >>And understanding is just a case of the person being able to attach the
> >>experience into his existing world-view in such way, that he thinks he
> >>has understood the subject matter. This thinking is also semantical. In
> >>a strict sense, he hasn't understood it. He has only interpetated it in
> >>a way that seems to make sense. He has placed "meaning" onto the experience.
> >
> > No. I can't tell you why here because it involves what I am currently
> > working on.
> >
> > There is a flaw in your logic above. You are unknowingly applying a
> > system to the one thing you can never successfuly apply it to. You
> > will always get answers that make sense to you but they will still be
> > incorrect. The problem is that most people never realize you would be
> > wrong.
>
> I'm not sure what you mean myself either, but perhaps I should clarify
> that the point was simply, that if you have already learned all kinds of
> things of the world, and you have reached consciousness in this process,
> then even if all your sensory input is cut, you will find yourself in
> the dark. There is nothing coming in through the sensory input to
> comprehend, but you still comprehend that you are not feeling, seeing,
> hearing or smelling anything.
Try to reach for what I am dangling here. I have already told you that
according to my theory, comprehension is a by-product of the
structuring of sensory input. You can comprehend nearly everything in
the exterior world. The problem comes when we try to use the process in
the incorrect fashion. Remeber that i claim sensory input is the basis
for comprehension. There is one thing we can never truly do. The
irony is that if we do cross that line, then we will still form a
theory which we can comprehend! It will be completely beleiveable but
entirely wrong at the same time. That is what has occured with
psychologists with regard to all available theories of mind. I am
stating that theories of mind are fundamentaly flawed because of one
single "assumption".
Regarding the loss of the sensory input in your comments above, if a
peson did suddenly loose all their senses, yes they will comprehend
that they have lost such. However, according to my theory, they will
never be able to truly comprehend what has caused such since all of the
sensory organs have been destroyed! It would be similar to falling
into a cognitve black hole of no return.
Yes, I should read up on that. Offhand, I have priorly been able to
explain away
all prior thing simply because of the illusion of mind. I call it "The
Decieving Consciousness".
".
I'll try my unique take on it here in a bit or so.
-Jason
Curt Welch
> I just started to read the book of Jeff Hawkins, called "On
> Intelligence". I haven't progressed very far, but it seems like he
> thinks just the way we do too. He is expressing frustration to "Turing
> cult" (well not in those words), and is working on some sort of AI
> system as well.
> And in fact, on the page that I am just on while writing this, he is
> talking about Chinese Room argument. And surprise surprise, he is
> stating basically what I stated;
>
> "As for me, I think Searle had it right. When I thought through the
> Chinese Room argument and when I thought about how computers worked, I
> didn't see understanding happening anywhere. I was convinced we needed
> to understand what "understanding" is, a way to define it that would
> make it clear when a system was intelligent and when it wasn't. It's
> behaviour doesn't tell us this."
But later on he flat out contradicts himself by saying that he believes
that it is possible to simulate the function of the neocortex with
computers and that if you did that, it would create intelligence and
understanding. There is nothing about the Chinese Room that would prevent
the rule books that the person is following from being a complete
simulation of a human brain. There is nothing about the Chinese Room that
prevents it from being a learning machine (as far as I know but to be
honest I'd have to see exactly how Searle defined it to be sure - as long
as they guy can write things down on paper and had a large supply of blank
paper, then it can become a learning machine).
The problem is that no one yet knows how to functionally describe
intelligence other than by pointing to the behavior of humans. And since
we don't know how to describe it, we have no way to know if a computer, or
a human acting like a computer by following rules and writing things down
on paper, could duplicate it's function or not.
What we do know about the Chinese Room is that a human attempting to act as
a computer would be so slow as to be totally impractical. The human in the
room would die long before they finished running the amount of code that
would be probably be required to duplicate intelligent human behavior long
enough to answer even a single question in Chinese. For example, if you
just attempted to estimate the number of neurons that would fire in a human
brain during the time it would take a human to answer a single question in
Chinese you would probably come up with a number in the billions. So a
computer simulation of the same processes would require the same billions
of neural transactions. And each neuron that fires might effect 10,000
other neurons, so to simulate that, each neuron that fires would require
the human to inspect the state of 10,000 other neurons by hand. So that's
trillions and trillions of manual operations the guy in the Chinese room
would have to follow just to answer one simple question. It's fairly easy
to believe that some simple process which takes a guy a few minutes to
perform using look tables is not something that would end up creating
intelligent human behavior. But can you honestly say that a human working
for 10 million years in the Chinese Room, following trillions of
instructions, would not be able to produce a human-like answer to a
question? The point is, we have no way to know because we don't know what
the brain is doing. But what we do know, is that for the human in the
Chinese room to duplicate what the brain is doing, it will take them
millions of years to do what the brain does in 5 minutes, which make the
thought experiment flat out unreasonable.
I too am reading Jeff's book right now and I'm about half way through it.
It struck me as very odd that he would say Searle got it right. But as I
read further, what I think he's actually saying is just that we need to
know what understanding is before we can look at something like the Chinese
Room and know if there is any understanding there or not.
And he also just seems to see the traditional approaches to AI as being the
wrong approach - yet he seems to be advocating a connectionist solution to
how the brain works, while at the same time, saying the the connectionist
approach is wrong. So he just seems to be contradicting himself right and
left in the book.
His main message however is that we need the correct framework for
understanding what intelligence is before we will be able to understand
what we are looking for in the brain, and before we will have any hope of
creating an intelligent machine. And I think everyone agrees with that
idea. His book then puts forth his best understanding of what the correct
framework is, and I think he's a lot closer than most - but I'm only half
way through the book.
Most his negative comments about other approaches is simply him saying they
got the conceptual framework wrong.
> -Anssi
Curt Welch
> Is this instruction language a universal language? If not there is no
> reason to suggest it can be used to interpret Chinese correctly. If so
> there is no proof such a language is in fact universal. And so far as
> I know Turing "language" only applies to computable numbers and
> there is no proof of equivalence between computable "language" in
> turing terms and human language in general.
Exactly.
That answer is the answer to AI - and that's the answer no one has. We
don't know the link between computers and intelligent human behavior yet
(or if there can be one). Without that answer, no one can say if the
Chinese Room has what it takes to be intelligent (and understand Chinese)
or not.
We do know enough about computers to prove that the Chinese room can do
anything a computer can - only slower. But we have nothing to tell us if
computers, if programmed correctly, could every understand in the way a
human understands.
All we can do is speculate and pick the side of the argument we like the
best.
Curt Welch
But your position is that we are a learning machine? How is a learning
machine not a databank? Where does it store what it has "learned" if not
in a "databank"???
Your position seems very self contradicting to me.
> It doesn't attack the chinese room argument so much as it attacks the
> idea it is based on; that semantics could be somehow intrinsic to our
> brain "like metabolism etc". Such an idea can only arise from an
> inability to see what else semantics could be.
>
> Well, since we do not actually know anything for certain, instead we
> only hold assumptions in our worldview, what else could arise but
> semantics? Everything connects in our worldview. And by connection I
> don't mean a low-level network connection, I mean higher abstraction, as
> to what meaning one thing has to another thing.
>
> (Obviously this stuff about semantics doesn't directly concern the
> low-level self-organization, such as your networks)
>
> I haven't come across similar counter-argument before. And it's pretty
> sad...
I think your counter argument is the systems argument and I see it as sad
that you don't realize that it's the same.
> > It all comes down to what you
> > choose to believe "understand" means. Some of us, believe that a
> > computer program has the power to "understand" and some of us do not.
>
> Sure, you just have to make the distinction between explicit
> understanding, such as what occurs in the brain at a low-level, or to
> your networks, and semantical understanding, such as the potential
> high-level behaviour of your network when trying to make sense of the
> world tby some method. That method for us is simply learning.
>
> > Only hard evidence that shows what understanding really is will give
> > anyone the power to debate the truth of the Chinese room. Until that
> > hard evidence shows up, everyone will continue to pick the side they
> > like best and stick to it.
>
> I think chinese room has its place in that, there are a lot of people
> working on AI systems who clearly don't have a clue. Being able to come
> over the chinese room argument; to understand what semantics have to do
> with intelligence, is very important. (ok, it may not be for your
> approach, but in principle it helps to understand this)
I see you using the word "semantics" but I don't see that you have defined
it in a mechanical way. Unless you can give us the mechanical description
of a machine which is intelligent, all you are doing is playing word games
by shifting the definition from one word to another.
The key undefined word connected with the Chinese room is "meaning". But
"understanding" and "knowledge" and "consciousness" and "concepts" and
"ideas" and "thoughts" are just a few of the thousands of words with share
the same confusion. In the end, no one knows how to define any of these
words in mechanical terms. And by "mechanical" I don't mean Lester's
mechanincal logic. I mean show me the design of a physical machine. Show
me the blue prints, or the software, or the chemical equations which
explain it's physical operation. Show me something physical that explains
what "meaning" is. Until you can do that, you have told us nothing about
the Chinese Room which we didn't already know.
The Chinese Room is something physical. It's a guy with a bunch of books
and paper and we assume, something to write with. That's enough to create a
learning machine, but yet we don't know if it's enough to create the same
type of learning machine which we believe humans to be. No one has yet
shown how to create a machine that can do what humans can do. Until we can
do that, no one, no matter what words they use, can correctly argue either
side of the Chinese Room debate.
> -Anssi
Lucky
Curt Welch wrote:
> In other words, you study it's behavior, and if it acts like a human, you
> know it's intelligent? How is this test not identical to the Turing test?
>
> --
> Curt Welch http://CurtWelch.Com/
> cu...@kcwc.com http://NewsReader.Com/
Curt,
1.) Assume Comprehension is a by-product of the brains sturcturing of
information. The brain has the ability to store information it
understands and information it does not understand. A diamond and a
lump of coal contain the same basic ingrediant, carbon. The differnce
between the two is how they are structured. I suggest a similar thing
is happening within the brain. Information understood is processed
differently from information not understood.
2.) I take it further. Your mind is simply the sum of that which you
comprehend. Accordingly, you can never think of anything other than
that wich you comprehend, even if only on the lowest of levels. Don't
believe me? Then think of something in which you possess absolutely no
comprehension of? What I am trying to get across is that which we
don't comprehend, is in itself, unconscious. So therefor comprehension
and conscious are the same exact thing.
3.) "Intelligence" is automaticly connected to "comprehension" in the
minds of people. Our current technology allows us to produce machines
that can store vast amounts of information, yet I have yet encountered
anyone who considers our current machines "Intelligent". Creating
machines that will simply mimic an intelligent responce is not the same
thing as comprehension. Nor will creating such machines produce the
type of future that A.I. researchers claim is "Just around the corner".
Any machine of this nature will not only be inferior to human
intelligence, but it will be inferior to that of animals as well.
Hell, dogs would be more advanced than any such machine.
4.) The only true test of "Intelligence" is a test for comprehension.
How do we test for comprehension in machines? The same way in which we
test comprehension skills in people. provide them a novel problem, one
that will require creativity to solve, mainly the forming of
connections between the current problem and problems encounterd in the
past. The machine will then have to choose amongst disimilar
situations from past experiences. This will require a complexity of
thought easily performed by humans, but only performed by those
machines that possess comprehension.
5.) How do we create A.I. with "Intelligence/Comprehension"?
Form a theory of Comprehension (I have based on the sturcutre of such
information)
Translate process of comprehension into programing
Resulting machines will mimick the process of comprehension, but the
comprehension itself will be completely genuine and exceptionaly
similar to humans.
The process appears to be full of generality but it actualy is not. I
can go into great depth as to how this is will be accomplished. What
will be produced is a "seed" intelligence/comprehension that the A.I.
itself will be responsible for expanding. If the time frame for this
expansion is similar to humans, then the resulting A.I. will procede
accordingly, maturing in a period of 20 or so years.
-Jason
Curt Welch
> Curt Welch wrote:
> > In other words, you study it's behavior, and if it acts like a human,
> > you know it's intelligent? How is this test not identical to the
> > Turing test?
> >
> > --
> > Curt Welch
> > http://CurtWelch.Com/ cu...@kcwc.com
> > http://NewsReader.Com/
>
> Curt,
>
> 1.) Assume Comprehension is a by-product of the brains sturcturing of
> information. The brain has the ability to store information it
> understands and information it does not understand. A diamond and a
> lump of coal contain the same basic ingrediant, carbon. The differnce
> between the two is how they are structured. I suggest a similar thing
> is happening within the brain. Information understood is processed
> differently from information not understood.
But either way, all you are saying here is that information is represented
by the physical structure of the brain without saying anything useful about
what that structure is. AI was founded 50 years ago on the belief that the
physical structure of the brain was how it stored information. There's
nothing new here.
> 2.) I take it further. Your mind is simply the sum of that which you
> comprehend. Accordingly, you can never think of anything other than
> that wich you comprehend, even if only on the lowest of levels. Don't
> believe me? Then think of something in which you possess absolutely no
> comprehension of? What I am trying to get across is that which we
> don't comprehend, is in itself, unconscious. So therefor comprehension
> and conscious are the same exact thing.
Ok, fine. But all you have said here is that we are consciously aware of
what we are aware of and that we are not aware of the things we are not
aware of. Isn't that all kinda obvious?
> 3.) "Intelligence" is automaticly connected to "comprehension" in the
> minds of people. Our current technology allows us to produce machines
> that can store vast amounts of information, yet I have yet encountered
> anyone who considers our current machines "Intelligent". Creating
> machines that will simply mimic an intelligent responce is not the same
> thing as comprehension.
But then below, you say just the opposite. You say the way to test for
intelligence is to ask it to perform a task and if it produces the correct
response you say it is intelligent. How would your test not be testing for
the machines ability to "mimic" an intelligent response?
> Nor will creating such machines produce the
> type of future that A.I. researchers claim is "Just around the corner".
> Any machine of this nature will not only be inferior to human
> intelligence, but it will be inferior to that of animals as well.
> Hell, dogs would be more advanced than any such machine.
You are once gain, saying nothing of substance here. You haven't defined
what intelligence is in a way that allows us to know if a given machine is
intelligent but yet you say that some machine which might be here tomorrow
can't be intelligent even thought 1) you haven't defined what it takes for
a machine to be intelligent is and 2) you haven't defined what machine will
be here tomorrow.
> 4.) The only true test of "Intelligence" is a test for comprehension.
> How do we test for comprehension in machines? The same way in which we
> test comprehension skills in people. provide them a novel problem, one
> that will require creativity to solve, mainly the forming of
> connections between the current problem and problems encounterd in the
> past. The machine will then have to choose amongst disimilar
> situations from past experiences. This will require a complexity of
> thought easily performed by humans, but only performed by those
> machines that possess comprehension.
That's called the Turning test. You ask the computer to "solve novel
problem X" and when it doesn't respond as you expect an intelligent agent
to respond, you say "that machine is not intelligent". Your test for
intelligence above is a limited subset of the general Turning test which
allows the testers to ask the computer anything they want to ask them and
see if the computer responds as they think an intelligent human would
respond. If it doesn't, then the computer fails the Turing test.
> 5.) How do we create A.I. with "Intelligence/Comprehension"?
>
> Form a theory of Comprehension (I have based on the sturcutre of such
> information)
>
> Translate process of comprehension into programing
Well, sure. All you have said here is, 1) form a theory of intelligence,
and 2) translate that theory into a machine design.
That's the definition of the field of AI.
> Resulting machines will mimick the process of comprehension, but the
> comprehension itself will be completely genuine and exceptionaly
> similar to humans.
Yeah, if 1) it's possible and 2) you manage to do it. But that's what
everyone is working on.
> The process appears to be full of generality but it actualy is not. I
> can go into great depth as to how this is will be accomplished. What
> will be produced is a "seed" intelligence/comprehension that the A.I.
> itself will be responsible for expanding. If the time frame for this
> expansion is similar to humans, then the resulting A.I. will procede
> accordingly, maturing in a period of 20 or so years.
Right, so you build a learning machine and let it learn. That's the theory
that Turning wrote about in his famous 1950 paper that defined the Turing
test. I hope you don't think that any of what you have shown us so far is
new.
If you actually have working code that allows a computer to learn at the
same rate a human learns, then that's something new. But the idea of
building a learning machine is as old as the field of AI. Lots of people
have worked on it but no one has yet built a machine with the same learning
skills that humans have.
I too happen to believe that we must build a learning machine to create AI,
and I happen to believe it will be possible for us to do this. But it's
all just speculation on my part.
> -Jason
Lester Zick
comp.ai.philosophy wrote:
There is no proof of equivalence between human language, behavior, and
computable numbers.
~v~~
Lucky
Curt Welch wrote:
> "Lucky" <Lucky...@aol.com> wrote:
> > Curt Welch wrote:
> > > In other words, you study it's behavior, and if it acts like a human,
> > > you know it's intelligent? How is this test not identical to the
> > > Turing test?
> > >
> > > --
> > > Curt Welch
> > > http://CurtWelch.Com/ cu...@kcwc.com
> > > http://NewsReader.Com/
> >
> > Curt,
> >
> > 1.) Assume Comprehension is a by-product of the brains sturcturing of
> > information. The brain has the ability to store information it
> > understands and information it does not understand. A diamond and a
> > lump of coal contain the same basic ingrediant, carbon. The differnce
> > between the two is how they are structured. I suggest a similar thing
> > is happening within the brain. Information understood is processed
> > differently from information not understood.
>
> But either way, all you are saying here is that information is represented
> by the physical structure of the brain without saying anything useful about
> what that structure is. AI was founded 50 years ago on the belief that the
> physical structure of the brain was how it stored information. There's
> nothing new here.
Curt, before we procede I would like you to switch the current
paradigm, frame of reference, that is unconsciously shaping your
perceptions, in this case of my statements.
I never said communication is a result of the way the brain is
structured. I stated comprehension is a result of the way in which
information is structured.
Going further, the brain has the ability to store two types of
information, information understood, and informaion not understood.
(understanding = comprehension)
My theory suggests "Information Understood" results from the processing
of information not understood into a structure. It's the structuring
procedure itself that produces comprehension.
Going further, a person cannot think of something they do not
comprehend. You will most likely not understand this and disagree, but
you will be wrong. There are levels of comprehension. If you
absolutely do not comprehend something in any form, then any such
information would be unconscious. Follow?
Therefore, "Thinking" = Comprehension. Since "thought" is a word with
connections to "Consciousness", I hereby state that that
"thinking/thought and "Consciousness" are synonyms of Comprehension.
>
> > 2.) I take it further. Your mind is simply the sum of that which you
> > comprehend. Accordingly, you can never think of anything other than
> > that wich you comprehend, even if only on the lowest of levels. Don't
> > believe me? Then think of something in which you possess absolutely no
> > comprehension of? What I am trying to get across is that which we
> > don't comprehend, is in itself, unconscious. So therefor comprehension
> > and conscious are the same exact thing.
>
> Ok, fine. But all you have said here is that we are consciously aware of
> what we are aware of and that we are not aware of the things we are not
> aware of. Isn't that all kinda obvious?
My comments above hopefully will clarify this further for you. You are
that which you understand!
>
> > 3.) "Intelligence" is automaticly connected to "comprehension" in the
> > minds of people. Our current technology allows us to produce machines
> > that can store vast amounts of information, yet I have yet encountered
> > anyone who considers our current machines "Intelligent". Creating
> > machines that will simply mimic an intelligent responce is not the same
> > thing as comprehension.
>
> But then below, you say just the opposite. You say the way to test for
> intelligence is to ask it to perform a task and if it produces the correct
> response you say it is intelligent. How would your test not be testing for
> the machines ability to "mimic" an intelligent response?
That's not what I stated at all. I said give them the exact same tests
that we give humans today to test their ability for comprehension. A
person who has poor comprehension is considered to be either
below-average intelligence or mentaly retarded. Thus there is indeed a
connection between intelligence and comprehension.
>
> > Nor will creating such machines produce the
> > type of future that A.I. researchers claim is "Just around the corner".
> > Any machine of this nature will not only be inferior to human
> > intelligence, but it will be inferior to that of animals as well.
> > Hell, dogs would be more advanced than any such machine.
>
> You are once gain, saying nothing of substance here. You haven't defined
> what intelligence is in a way that allows us to know if a given machine is
> intelligent but yet you say that some machine which might be here tomorrow
> can't be intelligent even thought 1) you haven't defined what it takes for
> a machine to be intelligent is and 2) you haven't defined what machine will
> be here tomorrow.
You are not comprehending my statements. Using my theory, it would be
due to you failing to apply the proper paradigm in order to process new
information.
1.) As I stated before, Intelligence is the ability to both store and
comprehend information.
2.) Since A.I. researchers just give up on trying to create
comprehension in A.I., then they will only succeed in creating vastly
intricate cumputer programs witch do not possess intellignece.
> > 4.) The only true test of "Intelligence" is a test for comprehension.
> > How do we test for comprehension in machines? The same way in which we
> > test comprehension skills in people. provide them a novel problem, one
> > that will require creativity to solve, mainly the forming of
> > connections between the current problem and problems encounterd in the
> > past. The machine will then have to choose amongst disimilar
> > situations from past experiences. This will require a complexity of
> > thought easily performed by humans, but only performed by those
> > machines that possess comprehension.
>
> That's called the Turning test. You ask the computer to "solve novel
> problem X" and when it doesn't respond as you expect an intelligent agent
> to respond, you say "that machine is not intelligent". Your test for
> intelligence above is a limited subset of the general Turning test which
> allows the testers to ask the computer anything they want to ask them and
> see if the computer responds as they think an intelligent human would
> respond. If it doesn't, then the computer fails the Turing test.
That's not Turing Test at all. Turing test is when a machine is able to
trick a person into thinking they are interacting with another person.
The key word is "trick", involving deception, involving con artist,
hooligan, snake oil salesman, etc.... A.I. using Turing Test is a
complete joke and fraud.
The tests given for comprehension are given to children to determine
their ability to apply infromation gathered from disimilar experiences
and apply them toward solving a new problem. They are quite simple for
most adults to perform. It is not a Turing test as each answer is
depends on the creativity of the person taking the test. A turing test
could never erase prior programs, create new programs by combining
parts from other program, or disobey their programming.
Think about this one... A machine that does not possess intelligence
could not commit "suicide" unless programmed too. Otherwise, an A.I.
with comprehension can freely choose to destroy itself and could
possibly leave a "suicide note" explaining it's decision and why it did
not wish to exist anymore.
>
> > 5.) How do we create A.I. with "Intelligence/Comprehension"?
> >
> > Form a theory of Comprehension (I have based on the sturcutre of such
> > information)
> >
> > Translate process of comprehension into programing
Already beat you to it.
>
> Well, sure. All you have said here is, 1) form a theory of intelligence,
> and 2) translate that theory into a machine design.
>
> That's the definition of the field of AI.
>
> > Resulting machines will mimick the process of comprehension, but the
> > comprehension itself will be completely genuine and exceptionaly
> > similar to humans.
>
> Yeah, if 1) it's possible and 2) you manage to do it. But that's what
> everyone is working on.
To the best of my knowledge, no one has formed a theory anywhere
similar to my own. (And I have asked around at lot in this field.)
Yes, eventualy I must put out or get out.
>
> > The process appears to be full of generality but it actualy is not. I
> > can go into great depth as to how this is will be accomplished. What
> > will be produced is a "seed" intelligence/comprehension that the A.I.
> > itself will be responsible for expanding. If the time frame for this
> > expansion is similar to humans, then the resulting A.I. will procede
> > accordingly, maturing in a period of 20 or so years.
>
> Right, so you build a learning machine and let it learn. That's the theory
> that Turning wrote about in his famous 1950 paper that defined the Turing
> test. I hope you don't think that any of what you have shown us so far is
> new
NOOOOOOO! I create a comprehending A.I., turn it on and let if LIVE!
It will
learn by living, the same as you and I.
.
>
> If you actually have working code that allows a computer to learn at the
> same rate a human learns, then that's something new. But the idea of
> building a learning machine is as old as the field of AI. Lots of people
> have worked on it but no one has yet built a machine with the same learning
> skills that humans have.
>
> I too happen to believe that we must build a learning machine to create AI,
> and I happen to believe it will be possible for us to do this. But it's
> all just speculation on my part.
If I were to share my theory with you, it would take you about 15
minutes of discussion before you realzied something, so I will just
spare you that experience. I have been cussed out by people at
Stanford and Harvard because they realized that if my theory is correct
and can produce comprehension in A.I. then whole careers have been a
wast and in vain. My theories do not utilize any curent proraming
available in A.I. research. My theories imply that we must literaly go
back to start and do this again.
-Jason
Curt Welch
Oh, I am very consciously aware that my current frame of reference is
shaping my perceptions. I'm happy to try and shift to your frame of
reference, but I must do it by hanging it on my frame. So far, I've not
found enough meat in your words to hang anything.
> I never said communication is a result of the way the brain is
> structured. I stated comprehension is a result of the way in which
> information is structured.
>
> Going further, the brain has the ability to store two types of
> information, information understood, and informaion not understood.
> (understanding = comprehension)
>
> My theory suggests "Information Understood" results from the processing
> of information not understood into a structure. It's the structuring
> procedure itself that produces comprehension.
>
> Going further, a person cannot think of something they do not
> comprehend. You will most likely not understand this and disagree,
I think I do agree and understand. But isn't it obvious? I do not
comprehend the French Language and I will never have thoughts about how
verbs work in the French language because I know nothing about how verbs
work in the French language. Isn't it obvious that we both don't (and
cannot) think about things we don't know anything about?
> but
> you will be wrong. There are levels of comprehension. If you
> absolutely do not comprehend something in any form, then any such
> information would be unconscious. Follow?
Not just unconscious, but outright not there at all.
> Therefore, "Thinking" = Comprehension. Since "thought" is a word with
> connections to "Consciousness", I hereby state that that
> "thinking/thought and "Consciousness" are synonyms of Comprehension.
Ok if you wish. I agree they are all closely associated, but typical usage
tends to indicate they are not synonyms. The activity which goes on in
our head and which we are aware of is what most people seem to refer to
when we talk about "thinking", (I had the thought that I would like to eat
ice cream), where as comprehension means that we understand the
implications of some data (I comprehend that the sign meant that I had to
stop).
> > > 2.) I take it further. Your mind is simply the sum of that which you
> > > comprehend. Accordingly, you can never think of anything other than
> > > that wich you comprehend, even if only on the lowest of levels.
> > > Don't believe me? Then think of something in which you possess
> > > absolutely no comprehension of? What I am trying to get across is
> > > that which we don't comprehend, is in itself, unconscious. So
> > > therefor comprehension and conscious are the same exact thing.
> >
> > Ok, fine. But all you have said here is that we are consciously aware
> > of what we are aware of and that we are not aware of the things we are
> > not aware of. Isn't that all kinda obvious?
>
> My comments above hopefully will clarify this further for you. You are
> that which you understand!
Sure, I agree with that. Or at least, my mind is that which I understand.
My body is more than just my mind and there is much about my body which I
do not understand but which I still call part of me.
I think most AI researchers would argue that they have never given up
trying to create machine comprehension. They might not know how to do it
and they might be working on some other AI project at the time, but that
doesn't mean they have given up on creating comprehension.
> then they will only succeed in creating vastly
> intricate cumputer programs witch do not possess intellignece.
>
> > > 4.) The only true test of "Intelligence" is a test for comprehension.
> > > How do we test for comprehension in machines? The same way in which
> > > we test comprehension skills in people. provide them a novel problem,
> > > one that will require creativity to solve, mainly the forming of
> > > connections between the current problem and problems encounterd in
> > > the past. The machine will then have to choose amongst disimilar
> > > situations from past experiences. This will require a complexity of
> > > thought easily performed by humans, but only performed by those
> > > machines that possess comprehension.
> >
> > That's called the Turning test. You ask the computer to "solve novel
> > problem X" and when it doesn't respond as you expect an intelligent
> > agent to respond, you say "that machine is not intelligent". Your test
> > for intelligence above is a limited subset of the general Turning test
> > which allows the testers to ask the computer anything they want to ask
> > them and see if the computer responds as they think an intelligent
> > human would respond. If it doesn't, then the computer fails the Turing
> > test.
>
> That's not Turing Test at all. Turing test is when a machine is able to
> trick a person into thinking they are interacting with another person.
> The key word is "trick", involving deception, involving con artist,
> hooligan, snake oil salesman, etc.... A.I. using Turing Test is a
> complete joke and fraud.
You are just being silly. The Turing test is not about tricks. It's about
producing behavior which is so close to human behavior that a human could
not tell if the behavior was created by a human or something else. It
includes being able to give the test subject any test you want to give it.
And there is no time limit on the test. So if you feel the need to test
the subject for 10 years, to see how it's memory works and to see how it
can learn and grow over the 10 years you test it, then that's still a valid
Turing test. You can spend 10 years teaching it to read French and
teaching it French history and then test it on it's ability to debate the
meaning of some event in history with experts in the field. It means you
can teach it how to do science and ask it to spend 100 years developing AI
and see what it comes up with. That's the Turning test.
The point of the Turning test (which BTW, Turning called "The imitation
game") was simply to show that the simplest way to define intelligence was
by external behavior as observed and understood by humans, and if you did
it that way, then there would be nothing left to debate about the meaning
of intelligence. Turning made up his imitation game as way to turn the
debate about intelligence away from silly arguments over the meaning of
words like "comprehension" and "understand" and convert it to a simple
objective test. He then went on to discuss what it would take to make a
computer pass the test. He did this so that he wouldn't have to waste any
of his time debating the meaning of the word "intelligence" with anyone and
could instead just study the far more interesting problem of making a
computer act like a human.
If you create a machine and give it test to see if it acts like a human
(has human comprehension), then you are giving your machine a Turning test
whether you understand it as such or not.
> The tests given for comprehension are given to children to determine
> their ability to apply infromation gathered from disimilar experiences
> and apply them toward solving a new problem. They are quite simple for
> most adults to perform. It is not a Turing test as each answer is
> depends on the creativity of the person taking the test. A turing test
> could never erase prior programs, create new programs by combining
> parts from other program, or disobey their programming.
Your English is getting real odd here. "A Turing test could never erase
programs"? What are you trying to say? You seem to be failing the Turing
test. :)
If you mean, "the computer being tested in the Turning test could never
erase programs", you are just wrong. A computer program that changes it's
program is called a learning machine. There are many different algorithms
for creating learning machines - and the one thing they all have in common
is that they change their programing when they learn - after they learn,
they act differently than before they learned something.
It is true that a computer can never disobey it's programing (ignoring the
issue of what happens when it "breaks"). But humans also are unable to
disobey their "programing" which is defined by the laws of nature in
combination with the physical structure of their body. A hydrogen atom can
never choose to act like an oxygen atom. It's forced to follow it's
programing and only act like a hydrogen atom. This is true for any and all
parts of the human body.
There's nothing a human does that can not be explained as the human body
just following its "programing". We can't easily prove that the human body
is forced to follow the laws of nature but it's a fairly safe bet to assume
it does.
> Think about this one... A machine that does not possess intelligence
> could not commit "suicide" unless programmed too.
Think about this one. A human can not commit suicide unless the laws of
nature (gravity, weak and strong atomic forces, etc), forced it to.
> Otherwise, an A.I.
> with comprehension can freely choose to destroy itself and could
> possibly leave a "suicide note" explaining it's decision and why it did
> not wish to exist anymore.
We already understand how to write computer programs that have the power to
decide on their own, when they want to commit suicide. See Microsoft
Windows and the blue screen of death. It even writes you a little suicide
note and it decides all on it's own when it wants to do that.
If you think a human committing suicide is doing something substantially
different, then I would argue you have no clue what AI and human
intelligence is all about. You are bestowing magic on us which just isn't
there.
> > > 5.) How do we create A.I. with "Intelligence/Comprehension"?
> > >
> > > Form a theory of Comprehension (I have based on the sturcutre of such
> > > information)
> > >
> > > Translate process of comprehension into programing
>
> Already beat you to it.
>
> >
> > Well, sure. All you have said here is, 1) form a theory of
> > intelligence, and 2) translate that theory into a machine design.
> >
> > That's the definition of the field of AI.
> >
> > > Resulting machines will mimick the process of comprehension, but the
> > > comprehension itself will be completely genuine and exceptionaly
> > > similar to humans.
> >
> > Yeah, if 1) it's possible and 2) you manage to do it. But that's what
> > everyone is working on.
>
> To the best of my knowledge, no one has formed a theory anywhere
> similar to my own. (And I have asked around at lot in this field.)
>
> Yes, eventualy I must put out or get out.
Well, I intend to stay around passing out my grandiose theories even if I
never mange to put out. :) But I also understand that unless I do put out,
I've got nothing to justify my talk with.
> > > The process appears to be full of generality but it actualy is not.
> > > I can go into great depth as to how this is will be accomplished.
> > > What will be produced is a "seed" intelligence/comprehension that the
> > > A.I. itself will be responsible for expanding. If the time frame for
> > > this expansion is similar to humans, then the resulting A.I. will
> > > procede accordingly, maturing in a period of 20 or so years.
> >
> > Right, so you build a learning machine and let it learn. That's the
> > theory that Turning wrote about in his famous 1950 paper that defined
> > the Turing test. I hope you don't think that any of what you have shown
> > us so far is new
>
> NOOOOOOO! I create a comprehending A.I., turn it on and let if LIVE!
> It will
> learn by living, the same as you and I.
Fine, so if we use the word "life" it's real AI, and if we use the word
"running machine" it's not AI?
> > If you actually have working code that allows a computer to learn at
> > the same rate a human learns, then that's something new. But the idea
> > of building a learning machine is as old as the field of AI. Lots of
> > people have worked on it but no one has yet built a machine with the
> > same learning skills that humans have.
> >
> > I too happen to believe that we must build a learning machine to create
> > AI, and I happen to believe it will be possible for us to do this. But
> > it's all just speculation on my part.
>
> If I were to share my theory with you, it would take you about 15
> minutes of discussion before you realzied something, so I will just
> spare you that experience.
Oh, I love AI theories - especial ones that are worth anything. I'll be
happy to spend 24 hours discussing it with you if that's what it takes.
> I have been cussed out by people at
> Stanford and Harvard because they realized that if my theory is correct
> and can produce comprehension in A.I. then whole careers have been a
> wast and in vain. My theories do not utilize any curent proraming
> available in A.I. research. My theories imply that we must literaly go
> back to start and do this again.
I think most people in AI believe no one yet has the full correct approach
to AI. AI is the search for that approach which will get us to the goal.
If any of the current systems were the correct approach, they would have
already led us to human level intelligent machines.
It's been said that AI is the only discipline in search for a definition
for itself.
But at the same time, it's highly likely that all previous AI work got some
part of it correct. We just won't know which part they got correct and
which they got wrong until someone finds the right answer.
Everyone that's worth anything knows we need a new approach and they are
already looking for it.
What you have written however seems to indicate that your major approach is
to use words like "comprehend" and "life" instead of words like "computer
software" and "Turing tests". I don't see that as a new approach. I see
that mostly as just ignoring the hard problem - which is how do you define
"life" and "comprehension" in terms of wheels, and levers, and transistors.
But maybe, you do have ideas about exactly how to do that and you just
don't want to share it yet.
1Z
Curt Welch wrote:
> But later on he flat out contradicts himself by saying that he believes
> that it is possible to simulate the function of the neocortex with
> computers and that if you did that, it would create intelligence and
> understanding. There is nothing about the Chinese Room that would prevent
> the rule books that the person is following from being a complete
> simulation of a human brain.
That depends on what you mean by simulation. It could be black-box
simulation that produces the same outputs tot the same inputs. It could
not be a white-box simulation (ie having the same functional
organsitation)
because it would be serial, not parallel.
Anssi Hyytiainen
> Anssi Hyytiainen <ans...@nic.fi.ANTISPAM> wrote:
>>explicit answers. We are learning machines. *We don't have any answers!*
>
> But your position is that we are a learning machine? How is a learning
> machine not a databank? Where does it store what it has "learned" if not
> in a "databank"???
It doesn't hold any explicit information. All it holds is its own
assumptions as to what something is, and how it connects to everything
else. It is not intelligent to look from a databank that a "bird" is "an
animal that flies". It is intelligent to figure out yourself how to
"see" the world.
> I think your counter argument is the systems argument and I see it as sad
> that you don't realize that it's the same.
System argument:
"Although the individual in the Chinese room does not understand
Chinese, perhaps the person and the room considered together as a system
do. Searle’s reply to this is that someone might in principle memorize
the rule book; they would then be able to interact as if they understood
Chinese, but would still just be following a set of rules, with no
understanding of the significance of the symbols they are manipulating."
Searle is right. Following a set of rules is not intelligent in itself.
Coming up with your very OWN idea of what a bird is, is intelligent.
> I see you using the word "semantics" but I don't see that you have
defined
> it in a mechanical way.
We cannot just "know" the meaning of anything. Bird is not explicitly an
animal that flies. That's just one way someone may choose to see it. We
are blinded by the naive intuition that there exists such things as
truths and falsehoods in our conscious understanding of the world.
Only "truths" are those that we declare ourself, such as the explicit
rules of math. The rules themself are just an algorithm, with no
intelligent of their own. That is to say, for a computer, the commands
are truths, and that's the end of it.
Mechanical semantics arise from such dynamic databanks that holds merely
a WAY to see the world, a WAY to interpetate. It's a methdo to tag
semantical meaning on things.
Whatever meaning one has tagged to something, it will affect how the
databank will grow in the future. If you think there is God, you will
continue to see God's handywork everywhere.
> I mean show me the design of a physical machine.
I can only give you a rough idea of its high-level logics. The logics
is, that when there's a stimulation from the outside world, the response
is not subject to an algorithm directly, rather it is subject to
algorithm and dynamic databank. The databank is not "the truth". It's
just "an idea". And it changes all the time.
> Show
> me the blue prints, or the software, or the chemical equations which
> explain it's physical operation. Show me something physical that
explains
> what "meaning" is.
Semantical meaning is simply something that has no explicit meaning.
Only assumed meaning.
> The Chinese Room is something physical. It's a guy with a bunch of books
> and paper and we assume, something to write with.
Sure, we just have to change the way the chinese room responds with mere
algorithm; the instruction table. Instruction table is not a dynamic
databank which shows a way to see things. It's just explicit set of
rules. A computer doesn't understand it's been fed with a bit "1" when
we do so. "1" doesn't have semantical meaning. A computer doesn't
understand any instruction is is being fed.
We need a machine that doesn't just follow a set of rules to come up
with behaviour. It does follow a set of rules to come up with "learning"
behaviour, but the way we learn is not a conscious process.
It needs to do its very own assumptions to put meaning on things. And
when it learns the semantical meaning of "I", it will be conscious.
Simple as that.
http://www.percepp.demon.co.uk/hkeller.htm
> That's enough to create a
> learning machine, but yet we don't know if it's enough to create the same
> type of learning machine which we believe humans to be. No one has yet
> shown how to create a machine that can do what humans can do. Until
we can
> do that, no one, no matter what words they use, can correctly argue
either
> side of the Chinese Room debate.
Chinese room is important argument in that it shows where the so-called
Turing cult is going wrong, and what to do to improve it.
-Anssi
Anssi Hyytiainen
> Semantics again, replace "learn" with "comprehend". Without
> comprehension
> then then a machine nor person, can recognize a problem to begin with.
Yes. Our semantically different ways to see consciousness have now
connected; we have the exact same idea. You need to comprehend the thing
in order for it to be a conscious experience. You cannot "comprehend"
the algorithmic part of the process with which one "learns", for example.
Great.
-Anssi
Anssi Hyytiainen
> Anssi Hyytiainen <ans...@nic.fi.ANTISPAM> wrote:
>>talking about Chinese Room argument. And surprise surprise, he is
>>stating basically what I stated;
>>
>>"As for me, I think Searle had it right. When I thought through the
>>Chinese Room argument and when I thought about how computers worked, I
>>didn't see understanding happening anywhere. I was convinced we needed
>>to understand what "understanding" is, a way to define it that would
>>make it clear when a system was intelligent and when it wasn't. It's
>>behaviour doesn't tell us this."
>
> But later on he flat out contradicts himself by saying that he believes
> that it is possible to simulate the function of the neocortex with
> computers and that if you did that, it would create intelligence and
> understanding.
I saw the same exact argument somewhere else on a book review, but it
just because you haven't quite grasped what he means. In fact, there is
no contradiction.
Turing had the right idea, but he was wrong to think that external
behaviour could equal to intelligence. Searle is right to point out that
it doesn't, but he then comes up to a wrong conclusion that semantics
are a fundamental property of brain. At heart of it all, Turing was in
fact RIGHT, but not the way he thought.
Not a single piece of the algorithmic syntax put into the computer can
be part of its conscious experience.
Conscious knowledge is not an algorithm. It is built BY an algorithm.
> There is nothing about the Chinese Room that would prevent
> the rule books that the person is following from being a complete
> simulation of a human brain.
It's still a rule book. If he is following something "blindly", he is
not drawing assumptions about anything.
> There is nothing about the Chinese Room that
> prevents it from being a learning machine (as far as I know but to be
> honest I'd have to see exactly how Searle defined it to be sure - as long
> as they guy can write things down on paper and had a large supply of
blank
> paper, then it can become a learning machine).
Yeah, that was the problem of the argument. Searle didn't allow learning.
> I too am reading Jeff's book right now and I'm about half way through it.
> It struck me as very odd that he would say Searle got it right. But as I
> read further, what I think he's actually saying is just that we need to
> know what understanding is before we can look at something like the
Chinese
> Room and know if there is any understanding there or not.
Well, I don't know what Jeff's idea of "understanding" is exactly
(although from short descriptions it sounds oddly familiar), but one
should really understand what is meant by Searle having a solid point in
his argument. Searle didn't put down the final word. He pointed out the
problem in the "puristic Turing approach".
Although if one understands Turing's point, one can also understand that
actually any mechanical approach is "puristic" Turing approach, and
that's why it would be wrong to say Turing was WRONG. He wasn't wrong in
implying that human mind is also mechanical. But he was wrong in saying
that external behaviour equals intelligence.
That's btw why that old thread of mine was called STRUCTURE of a
self-conscious mind. It is imperative to understand that it is the
specific structure of the mechanics that springs such a process that can
become conscious.
So, it didn't struck me as odd at all that he pointed out this thing
about Chinese Room. I was almost expecting him to, one way or another.
> And he also just seems to see the traditional approaches to AI as
being the
> wrong approach - yet he seems to be advocating a connectionist
solution to
> how the brain works, while at the same time, saying the the connectionist
> approach is wrong. So he just seems to be contradicting himself
right and
> left in the book.
He just sees the whole thing from a different angle than you, that's all.
How do I know this without even reading the book, well, different ways
to see the same thing is always the reason why any assertions may seems
contradicting to someone in the first place. That's not to say anything
about whose idea is more correct. After all, "all the ideas are wrong,
or imperfect at best", which is also one of the main points of what
semantics is.
> His main message however is that we need the correct framework for
> understanding what intelligence is before we will be able to understand
> what we are looking for in the brain, and before we will have any hope of
> creating an intelligent machine. And I think everyone agrees with that
> idea.
Yeah, with one small exception that I've stated before. I believe the
first conscious machine will be built without applying our own
"intelligent design"; it will be brewed in an evolutionary simulation.
The only "artificial" systems that could be said to have any traces of
real "intelligence" in them currently are, ironically, not intelligently
designed. They evolve. Like in Avida.
-Anssi
Anssi Hyytiainen
Behaviour is only thing we can study. I was pointing out that
self-conscious machines become spontaneously worried of their fate. Or
possibly the fate of the others.
-Anssi
Anssi Hyytiainen
> "Anssi Hyytiainen" <ans...@nic.fi.ANTISPAM> wrote in message
> news:T55If.6791$oD5....@reader1.news.jippii.net...
>
> Is this mechanical learning a hardwired instinct or an instinct who's
> strength can be adjusted by experience at critical stages?
Well, the important point of that assertion is simply that while
everything we consciously learn ends up as a "semantical declaration"
into our conscious mind, the learning process itself is not conscious;
it is just algorithmic.
And that is why "semantics" too, are at heart mechanical, although a
mechanical process *itself* is never semantical -> conscious.
When we consciously understand we have "learned" something, that is not
a case of "mechanical learning". The mechanical learning refers to just
the purely physical response that happens in our brains to the sensory
stimuli. We are not aware of "placing a concept into our association
network". "Association network", and "worldview" are purely semantical
concepts, dreamed up to offer a framework to discuss these ideas. There
are many ways to see these things.
Perhaps "learning" is not the best of all words to describe the
mechanical part of it, but I can't think of anything better to make this
point. Does that seem to click?
(To answer your question, it doesn't matter if the learning process is
hardwired or changes during our life, but I wish not to make any
assertion about our "instinct to learn". That I guess is just almost the
same thing as "curiosity")
-Anssi
Anssi Hyytiainen
> On Mon, 13 Feb 2006 21:24:10 +0200, Anssi Hyytiainen
> <ans...@nic.fi.ANTISPAM> in comp.ai.philosophy wrote:
>
>
>>Chinese Room argument goes as follows:
>>---
>>Imagine a native English speaker who knows no Chinese locked in a room
>>full of boxes of Chinese symbols together with a book of instructions
>>for manipulating the symbols. Imagine that people outside the room send
>>in other Chinese symbols which, unknown to the person in the room, are
>>questions in Chinese. And imagine that by following the instructions in
>>the program the man in the room is able to pass out Chinese symbols
>>which are correct answers to the questions. The program enables the
>>person in the room to pass the Turing Test for understanding Chinese but
>>he does not understand a word of Chinese.
>
>
> It occurs to me there is an implicit assumption here which needs to be
> clarified. In what language are the instructions written?
>
> Obviously they can't be written in Chinese because the English speaker
> doesn't understand Chinese. So saying the instructions are written in
> English or any other language understood by the English speaker and
> gives uniformly correct answers to questions in Chinese means Chinese
> is exhaustively subsumed by some other language.
It's not the point how the instruction table is written. It can be just
a picture of input symbols with arrows pointing to appropriate output
symbols. The point is that the response of the chinese room is
mechanical, algorithmic, computational, whatever you want to call it.
If you know about the mechanical manner with which computers manipulate
information, you can see that Searle is simply describing such a
process. It doesn't matter how fast or how "parallel" this information
manipulation is. If it is just about following an algorithm, it is not
intelligent.
Let's not think that when a complex algorithm can take all sorts of
pre-programmed things into account when coming up with the correct
response, it would be somehow intelligent. It is not. It doesn't know
any meaning of anything. It doesn't think.
It's just like a water flowing down more and more complex hills, still
ending up just where it was "supposed" to end.
The structure of such "AI" is wrong. The structure needs to be one that
doesn't follow rules to respond. It needs to learn things in a
semantical manner. It needs to be error-prone in its conclusions, it
needs to guess, it needs to hold "ideas".
-Anssi
Lester Zick
<ans...@nic.fi.ANTISPAM> in comp.ai.philosophy wrote:
>Lester Zick wrote:
> > On Mon, 13 Feb 2006 21:24:10 +0200, Anssi Hyytiainen
> > <ans...@nic.fi.ANTISPAM> in comp.ai.philosophy wrote:
> >
> >
> >>Chinese Room argument goes as follows:
> >>---
> >>Imagine a native English speaker who knows no Chinese locked in a room
> >>full of boxes of Chinese symbols together with a book of instructions
> >>for manipulating the symbols. Imagine that people outside the room send
> >>in other Chinese symbols which, unknown to the person in the room, are
> >>questions in Chinese. And imagine that by following the instructions in
> >>the program the man in the room is able to pass out Chinese symbols
> >>which are correct answers to the questions. The program enables the
> >>person in the room to pass the Turing Test for understanding Chinese but
> >>he does not understand a word of Chinese.
> >
> >
> > It occurs to me there is an implicit assumption here which needs to be
> > clarified. In what language are the instructions written?
> >
> > Obviously they can't be written in Chinese because the English speaker
> > doesn't understand Chinese. So saying the instructions are written in
> > English or any other language understood by the English speaker and
> > gives uniformly correct answers to questions in Chinese means Chinese
> > is exhaustively subsumed by some other language.
>
>It's not the point how the instruction table is written.
It sure as hell does matter. Obviously it's not the "point" of the
problem but all implicit assumptions about the problem depend
exactly on the assumption that there can be such an instruction
language and its nature.
We know the instruction language cannot be Chinese but could it be
English for example? If we are to assume that correct answers to
questions in Chinese are given in Chinese by a non Chinese speaker
according to the instruction language in English, that would make
Chinese merely a subset of English which I imagine would irritate
and amaze comparative philologists no end.
Below you suggest the instruction language could be just a set of
pictures and arrows. But the question is really could it be? Let's
suppose we had such a translation grid. The problem then becomes not
so much whether it could produce answers but whether it can produce
correct answers. And what you'll find is that the kind of answers you
can get and the kind of significance which can attach to the symbols
depends exactly on the instruction language.
Previously you complained symbols are only symbols and cognition is
more than mere symbols. And that's true. However what makes that true
is the kind of instruction language for translation and manipulation.
Hell even I could give a transliteration in Chinese but that wouldn't
make the answers correct to a native Chinese speaker. Nor would it
satisfy turing criteria for intelligence.
> It can be just
>a picture of input symbols with arrows pointing to appropriate output
>symbols. The point is that the response of the chinese room is
>mechanical, algorithmic, computational, whatever you want to call it.
And how exactly are ordinary translation factors non mechanical and
non algorithmic exactly? You use these terms as if they were de facto
pejorative without showing why and how they are pejorative.
>If you know about the mechanical manner with which computers manipulate
>information, you can see that Searle is simply describing such a
>process. It doesn't matter how fast or how "parallel" this information
>manipulation is. If it is just about following an algorithm, it is not
>intelligent.
So sez you. But I don't see you producing any instructions which are
intelligent or even explaining how and why they're intelligent and
mere instruction sequences not intelligent.
>Let's not think that when a complex algorithm can take all sorts of
>pre-programmed things into account when coming up with the correct
>response, it would be somehow intelligent. It is not.
I'm not assuming anything. You're assuming the reverse by asserting
the Chinese puzzle instructions could be just anything when in point
of fact what you'll find is that the quality of instruction is exactly
what governs the quality, significance, and intelligence of results.
Garbage in, garbage out unless instruction sequences are non trivial.
> It doesn't know
>any meaning of anything. It doesn't think.
Yeah sure except you don't know the meaning of meaning or thinking.
>It's just like a water flowing down more and more complex hills, still
>ending up just where it was "supposed" to end.
Whatever.
>The structure of such "AI" is wrong.
Well the structure of such ai may in fact be wrong but that doesn't
make your analysis of the problem right or even accurate. Certainly
your analysis of the Chinese Room Puzzle is fundamentally flawed.
> The structure needs to be one that
>doesn't follow rules to respond.
You just mean it needs to follow your rules instead of someone else's.
> It needs to learn things in a
>semantical manner. It needs to be error-prone in its conclusions, it
>needs to guess, it needs to hold "ideas".
Sure it does. But that just means the mechanics involved are at issue.
The only relevant scientific question for ai is whether intelligence,
cognition, and human predicate language represent computable numbers.
If so turing mechanics prevails, if not turning mechanics fails. And
the same people who are fooled by computer programs such that they
cannot say the programs are not intelligent are the same idiots who
would be fooled by any sequences of digits from the square root of two
or pi or any other non repeating decimal sequences.
~v~~
Curt Welch
> Curt Welch wrote:
> > Anssi Hyytiainen <ans...@nic.fi.ANTISPAM> wrote:
> >>And in fact, on the page that I am just on while writing this, he is
> >>talking about Chinese Room argument. And surprise surprise, he is
> >>stating basically what I stated;
> >>
> >>"As for me, I think Searle had it right. When I thought through the
> >>Chinese Room argument and when I thought about how computers worked, I
> >>didn't see understanding happening anywhere. I was convinced we needed
> >>to understand what "understanding" is, a way to define it that would
> >>make it clear when a system was intelligent and when it wasn't. It's
> >>behaviour doesn't tell us this."
> >
> > But later on he flat out contradicts himself by saying that he
> > believes that it is possible to simulate the function of the neocortex
> > with computers and that if you did that, it would create intelligence
> > and understanding.
>
> I saw the same exact argument somewhere else on a book review, but it
> just because you haven't quite grasped what he means. In fact, there is
> no contradiction.
>
> Turing had the right idea, but he was wrong to think that external
> behaviour could equal to intelligence.
I actually agree with most the basic ideas you keep putting forth. I just
believe you are contradicting yourself without realizing it in a few
points.
The point about external behavior is that there's now way in hell you could
get the external behavior correct, without getting the internal behavior
correct. You can for example make a puppet look intelligent for about 5
seconds just by making it mimic intelligent actions. But to make it look
intelligent after 30 days of testing, you have to get all the internal
functions correct. It has to learn and understand exactly like humans
learn and understand or else the external behavior will not look
intelligent after extended testing. Most important, none of us knows what
is going on inside someone else's body but yet we have no problem seeing
intelligence in their behavior. So it's a simple and obvious fact that the
_only_ way any of us currently spots intelligence in others is by observing
their external behavior. To not understand this is to be blind to the
obvious.
If you think Turning got something wrong, you don't understand how much you
still have wrong.
Curt Welch
Sure. But we know it's worried about these things only because of it's
external behavior. And if we didn't see this same external behavior in a
machine, we would not believe it had the same type of consciousness we
have.
Everything we understand about intelligence in others is based on what we
see happening with their external beahvior. If we saw _all_ the same
behaviors in a machine, would we be forced to conclude it was just as
conscious as the humans which we see acting the same way.
Curt Welch
> The structure of such "AI" is wrong. The structure needs to be one that
> doesn't follow rules to respond. It needs to learn things in a
> semantical manner. It needs to be error-prone in its conclusions, it
> needs to guess, it needs to hold "ideas".
Yes, I agree with all that. Except, to solve AI, you have to figure out
how to do that with a fixed machine design where the machine is both
following it's programing, but yet at the same time, learning in a
semantic manner and being error-prone and making guesses.
This seems like a contradiction to some, but it's not. We already know how
to make computers do these things. We just don't yet know how to make them
do it at human levels of performance.
Immortalist
This principle of your probably would work most of the time but I can
imagine a situation where things would get so tangled and the people
conversing so similar that the principle would have many exceptional reasons
for being denied a universal extension to all posting by all people. I
believe the poster was out of line and offered no reasons for the
exceptional behavior.
> -Jason
>
Immortalist
"Anssi Hyytiainen" <ans...@nic.fi.ANTISPAM> wrote in message
> Immortalist wrote:
> > "Anssi Hyytiainen" <ans...@nic.fi.ANTISPAM> wrote in message
> > news:T55If.6791$oD5....@reader1.news.jippii.net...
> >>Semantical understanding is due to mechanical learning:
> >
> > Is this mechanical learning a hardwired instinct or an instinct who's
> > strength can be adjusted by experience at critical stages?
>
> Well, the important point of that assertion is simply that while
> everything we consciously learn ends up as a "semantical declaration" into
> our conscious mind, the learning process itself is not conscious; it is
> just algorithmic.
>
So your claiming we can ignore anything below the semantical declaration?
Pretty good ploy actually. I agree then that some hardwired neural capacity
allows the mechanical learning. On that;
...from a Martian's-eye-view, all humans speak a single language is based on
the discovery that the same symbol- manipulating machinery, underlies the
world's languages. Universal Grammar is like an archetypal body plan found
across vast numbers of animals in a phylum (238-9), a common plan of
syntactic, morphological, and phonological rules and principles, with a
small set of varying parameters. Once set, a parameter can produce
far-reaching changes in the superficial appearance of the language. One of
the most intriguing discoveries is that there appears to be a common anatomy
in all phrases in all the world's languages. Phrase structure is the kind of
stuff language is made of; traces, cases, X-bars, and the other
paraphernalia of syntax are colourless, odourless, and tasteless, but they,
or something like them, must be a part of our unconscious mental life (124).
The universal plan underlying languages, with auxiliaries and inversion
rules, nouns and verbs, subjects and objects, phrases and clauses, cases and
agreement, and so on, seem to suggest a commonality in the brains of
speakers, because many other plans would have been just as useful (43).
3.2 Chomsky and the acquisition of language by children
Children must innately be equipped with a plan common to the grammars of all
language, a Universal Grammar, that tells them how to distil the syntactic
patterns out of the speech of their parents. The unordered super-rules
(principles) are universal and innate; when children learn a particular
language, they do not have to learn a long list of rules, because they are
born knowing the super-rules (112). All they have to learn is whether their
particular language has the parameter head-first, as in English, or
head-last, as in Japanese. If the verb comes before the object, the child
concludes that the language is head-first as if the child were merely
flipping a switch to one of two possible positions. The way language works
is that each person's brain contains a lexicon of words and the concepts
they stand for (a mental dictionary) and a set of rules that combine the
words to convey relationships among concepts (a mental grammar) (85).
Pinker suggests that language should be considered as an evolutionary
adaptation like the eye, its major parts designed to carry out important
functions. Chomsky speaks about the 'language organ' and language is assumed
to be a distinct brain module. The evolution of the eye was a central
debating point between proponents and opponents of Darwinian natural
selection; Darwin himself offered a gradualistic account which has been
taken up and refined many times, most notably recently by Richard Dawkins
(1986). A plausible account has been given of how even a rudimentary eye
could be selected and increase the fitness of the individual in whom the
advance took place; each small improvement in the functioning of the eye
would promote survival of the individual and the individual's offspring
carrying the gene for the improved eye. Maynard Smith and Szathmary in their
recent The Major Transitions in Evolution (1995) also pick up the analogy
between development of the eye and the development of language. This is
perhaps not surprising since admittedly they relied heavily on two sources,
Bickerton's (1990) book and Pinker and Bloom's 1990 article which
foreshadowed The Language Instinct.
3.4.2 The analogy with the evolution of instincts
Language is a distinct piece of the biological makeup of the brain (18). The
universality of complex language is the first reason to suspect that
language is the product of a special human instinct. If language is an
instinct, it should have an identifiable seat in the brain, and perhaps even
a special set of genes that help wire it into place (45-46). If language is
like other instincts, presumably it evolved by natural selection, the only
successful scientific explanation of complex biological traits (354), the
only alternative (Pinker's emphasis) that can explain the evolution of a
complex organ like the eye or like language. Each step in the evolution of a
language instinct, up to and including the most recent ones, must enhance
fitness (366- 367), the gradual accumulation over generations of random
genetic mutations that increase reproductive success (333). The language
instinct is composed of many parts: syntax, with its discrete combinatorial
system building phrase structures; morphology, a second combinatorial system
building words, a capacious lexicon; a revamped vocal tract; phonological
rules and structures; speech perception; parsing algorithms, learning
algorithms (362).
http://www.percepp.demon.co.uk/pinker.htm
> And that is why "semantics" too, are at heart mechanical, although a
> mechanical process *itself* is never semantical -> conscious.
>
Neither is seeing or hearing or anything outside of the activities of the
necessary neural behavior which is consciousness. Who claimed consciousness
doesn't require particular activities?
> When we consciously understand we have "learned" something, that is not a
> case of "mechanical learning". The mechanical learning refers to just the
> purely physical response that happens in our brains to the sensory
> stimuli. We are not aware of "placing a concept into our association
> network". "Association network", and "worldview" are purely semantical
> concepts, dreamed up to offer a framework to discuss these ideas. There
> are many ways to see these things.
>
> Perhaps "learning" is not the best of all words to describe the mechanical
> part of it, but I can't think of anything better to make this point. Does
> that seem to click?
>
> (To answer your question, it doesn't matter if the learning process is
> hardwired or changes during our life, but I wish not to make any assertion
> about our "instinct to learn". That I guess is just almost the same thing
> as "curiosity")
>
All humans talk but no house pets or house plants do, no matter how
pampered, so heredity must be involved in language. But a child growing up
in Japan speaks Japanese whereas the same child brought up in California
would speak English, so the environment is also crucial. Thus there is no
question about whether heredity or environment is involved in language, or
even whether one or the other is "more important." Instead, language
acquisition might be our best hope of finding out how heredity and
environment interact. We know that adult language is intricately complex,
and we know that children become adults. Therefore something in the child's
mind must be capable of attaining that complexity. Any theory that posits
too little innate structure, so that its hypothetical child ends up speaking
something less than a real language, must be false. The same is true for any
theory that posits too much innate structure, so that the hypothetical child
can acquire English but not, say, Bantu or Vietnamese.
http://www.ecs.soton.ac.uk/~harnad/Papers/Py104/pinker.langacq.html
> -Anssi
Lucky
Lester Zick wrote:
> Below you suggest the instruction language could be just a set of
> pictures and arrows. But the question is really could it be? Let's
> suppose we had such a translation grid. The problem then becomes not
> so much whether it could produce answers but whether it can produce
> correct answers. And what you'll find is that the kind of answers you
> can get and the kind of significance which can attach to the symbols
> depends exactly on the instruction language.
You are not understanding that the "The Chinese Room" Arguement is
flawed
for one thing, and one thing only.
I explained that when first attempting to understand the arguement we
must first
discover if any faulty logic was used in it's creation.
Unfortunatley "The Chines Room" Arguement does contain faulty logic.
It depends on system created of individual units, in this case people,
who
themselves possess Intelligence.
Therefore, human intelligence can be utililysed to overcome any
obstacle implied in the arguement itself. A human being can therefore
follow the instructions and correctly produce the corresponding Chinese
Character without any prior knowledge of the Chinese Language.
Going further, since Humans posess the ability to learn by repetition,
experience/involvement, and context, the man in the room might be able
to form a "self-taught" form of Chinese that in itself would pass as if
he truly learned the language in a more "normal" fashion!
> Previously you complained symbols are only symbols and cognition is
> more than mere symbols. And that's true. However what makes that true
> is the kind of instruction language for translation and manipulation.
> Hell even I could give a transliteration in Chinese but that wouldn't
> make the answers correct to a native Chinese speaker. Nor would it
> satisfy turing criteria for intelligence.
Yes, but the reasons why this is so is due to an inherent flaw in human
development. In order to properly understand chinese, all prior
knowledge of any other language would need to be compartmentalized or
somehow destroyed. The individual would then have to devote the same
number of years, from infancy forward, to learning Chinese. If any
other method is used, the brain has no choice but to use the primary
language as a paradigm to mold the understanding of the secondary
language.
Going further, The brain has the ability to swap primary and secondary
languages. For example, if you took classes in French and then moved
to France permanently, you would for a time still think in English and
then translate that into French before speaking. However, in time this
will change. From what I have learned from communicating with such
people, is that their dreams begin to change. They start dreaming in
the other language! Once they begin dreaming in the other language
they then notice they are thinking in the other language. Therefore,
French which was once the secondary language has now been switched to
become the primary language!
Note: I previously theorized that Dreams play a vital role in the
process of comprehension/understanding.
-Jason
Lucky
>self-conscious machines become spontaneously worried of their fate. Or
>possibly the fate of the others.
>-Anssi
Is Behavior the only thing we can study? Talk to me more about this.
I already have some "deep thoughts" on this.
-Jason
Lucky
>pampered, so heredity must be involved in language. But a child growing up
>in Japan speaks Japanese whereas the same child brought up in California
>would speak English, so the environment is also crucial.
Uh....Not in your analogy. Humans posess the capacity for Language
period.
It doesn't matter if it's Japanese, English, or any other spoken
Language for that
matter. All spoken languages use the medium of the vocal chords,
throat, mouth,
and tongue to create and control sounds. Therefore any human can learn
any
spoken language.
> Thus there is no
>question about whether heredity or environment is involved in language, or
>even whether one or the other is "more important."
Incorrect. Yes we do know of cases where "environment" is vital to the
development of language. If a small child is kept locked up, and not
able to both hear language being used, then the child will never be
able to properly grasp Language when older. This is because of a
processes called "Neural Reinforcement", as well as "Plasticity" of the
brain. If the neural pathways dedicated to spoken language are not
stimulated within the "window of opportunity" allotted/provided to
children, then the brain will divert the pathways and assign them
different tasks.
> Instead, language
>acquisition might be our best hope of finding out how heredity and
>environment interact. We know that adult language is intricately complex,
>and we know that children become adults. Therefore something in the child's
>mind must be capable of attaining that complexity. Any theory that posits
>too little innate structure, so that its hypothetical child ends up speaking
>something less than a real language, must be false. The same is true for any
>theory that posits too much innate structure, so that the hypothetical child
>can acquire English but not, say, Bantu or Vietnamese.
Wow...I don't know how to respond to this but suggest you do more
research about the developing brain, the acquisition of language, and
the structuring of neurons. Also do check into "Neural Reinforcement"
as well as "Plasticity" of the brain. I wouldn't comment further on
this line until you possess some basic information.
-Jason
Immortalist
Actually there are mental activities also, the time it takes to think up
solutions to problems.
behaviourism
http://www.ship.edu/%7Ecgboeree/beh.html
http://xrefer.com/entry.jsp?xrefid=551414
A family of doctrines united by metaphysical worries about dualism and
epistemic worries about the status of mental terms (even when not
undergirded by a dualistic metaphysic). Operationalism, positivism, and
behaviourism were mutually inspiring doctrines designed, in the case of
psychology, to make it scientifically respectable. Psychology, traditionally
conceived as the science of mind, became conceived as the science of
behaviour, where behaviour was understood to include only the 'observable'
activities of an organism, or, in the version B. F. Skinner dubbed 'radical
behaviourism', where behaviour was conceived of expansively so that 'private
events' like thinking, feeling, and so on, although not directly observable
were taken to be kinds of behaviour subject to the same laws as more public,
conspicuous behaviour. Every type of behaviourism involved some sort of
challenge to 'mental realism', to our ordinary way(s) of thinking of mind
and mentality. Some of the more interesting behaviouristic doctrines include
the following:
Operationalistic behaviourism
The meaning of a mental term is exhausted by the observable operations that
determine its use. So 'P is thirsty' means P says she is thirsty if asked,
drinks water if given the chance, and so on.
Logical behaviourism
Mental terms are disposition terms. To say that 'P is thirsty' is to imply,
among other things, that P will probably say she is thirsty if asked, will
drink if given the chance, and so on. The difference between the first and
second doctrine is that the first denies any 'surplus meaning' to the
concept of 'thirst' beyond that entailed by the observations used in the
determination to use it; whereas the second allows that the concept of
'thirst' is only partially reduced to the observable events that justify its
use, and thus that it maintains a legitimate surplus meaning referring to a
'state' inside the organism, the qualitative state, say, of 'being thirsty'.
Methodological behaviourism
Despite the fact that there are private psychological events, 'psychology',
conceived as the science of behaviour, can avoid talking about them, and
thereby retain its scientific credentials. The basic idea was pointed out by
B. F. Skinner in Science and Human Behavior (1953) and was picked up and
elaborated on by Carl Hempel, who called it the 'Theoretician's Dilemma'.
Assuming that unobservable private events serve to link stimuli and
responses in lawlike ways, we can, for purposes of psychology, treat the
mind as a black box, observing the effects of the environment on behaviour,
and predicting and explaining behaviour on that basis.
Radical behaviourism
The doctrine that behaviour can be observable or unobservable (from the
third-person point of view) but that both can be analysed within the
substantive framework of behaviouristic psychology. In 'Behaviorism at 50'
(1964), Skinner writes: 'It is especially important that a science of
behavior face the problem of privacy. It may do so without abandoning the
basic position of behaviorism. Science often talks about things it cannot
see or measure ... The skin is not that important a boundary.' With the
advent of radical behaviourism, one sees the attempt on Skinner's part to
argue for the thesis that all behaviour, public or private, is governed by
the laws of classical conditioning (as articulated by Pavlov and Watson) or
operant conditioning (as articulated by Thorndike and himself). Skinner
argued that thinking, choosing, and deciding - things about which more
draconian forms of behaviourism vowed silence - could be analysed as private
behaviours with characteristic causal relations to overt behaviour and as
subject to the basic principles of operant conditioning. Despite this
expansiveness, Skinner remained unimpressed until his dying day with the
rising tide of cognitive psychology, thinking it lacked epistemic discipline
and was rudely ignorant of the contributions of the substantive doctrines of
classical and operant behaviourism. Although no version of behaviourism is a
live position within the philosophy of mind, most philosophers still think
that mental terms typically get at least part of their meaning from links to
observable causes and effects.
> -Jason
>
Lucky
Curt Welch wrote:
> > Curt, before we procede I would like you to switch the current
> > paradigm, frame of reference, that is unconsciously shaping your
> > perceptions, in this case of my statements.
>
> Oh, I am very consciously aware that my current frame of reference is
> shaping my perceptions. I'm happy to try and shift to your frame of
> reference, but I must do it by hanging it on my frame. So far, I've not
> found enough meat in your words to hang anything.
LOL, Curt if I explained my theory to you would see how ridiculous the
above comments sound. You are assuming you have a choice in
comprehending. I am saying you don't. :)
> > I never said communication is a result of the way the brain is
> > structured. I stated comprehension is a result of the way in which
> > information is structured.
> >
> > Going further, the brain has the ability to store two types of
> > information, information understood, and informaion not understood.
> > (understanding = comprehension)
> >
> > My theory suggests "Information Understood" results from the processing
> > of information not understood into a structure. It's the structuring
> > procedure itself that produces comprehension.
> >
> > Going further, a person cannot think of something they do not
> > comprehend. You will most likely not understand this and disagree,
>
> I think I do agree and understand. But isn't it obvious? I do not
> comprehend the French Language and I will never have thoughts about how
> verbs work in the French language because I know nothing about how verbs
> work in the French language. Isn't it obvious that we both don't (and
> cannot) think about things we don't know anything about?
LOL. You know that French is a language, therefore you possess some
comprehension of it. :)
> > but
> > you will be wrong. There are levels of comprehension. If you
> > absolutely do not comprehend something in any form, then any such
> > information would be unconscious. Follow?
>
> Not just unconscious, but outright not there at all.
LOL, how many times have I written that the brain has the ability to
store information it understands and information it doesn't understand.
I have also previously stated that what we call "conscious" is simply
made up of that which we comprehend, so "consciousness =
comprehension". If you don't comprehend something, and thus aren't
conscious of it, you have no idea if it exists in random bits of
information stored in your brain which have yet to be processed into
comprehension!
> > Therefore, "Thinking" = Comprehension. Since "thought" is a word with
> > connections to "Consciousness", I hereby state that that
> > "thinking/thought and "Consciousness" are synonyms of Comprehension.
>
> Ok if you wish. I agree they are all closely associated, but typical usage
> tends to indicate they are not synonyms.
I agree. :) That's why I just proved to you that these "disimilar"
things are actualy aspects of the same process. Hint: let go of the
illusion that you control the process of comprehension, because you
dont. The only thing you can do is to cling to tightly to a paradigm
preventing further comprehension. Let go of your paradigms. Surrender
to the process. The waters of comprehension are not deep and I assure
you that you will not drown.
> The activity which goes on in
> our head and which we are aware of is what most people seem to refer to
> when we talk about "thinking", (I had the thought that I would like to eat
> ice cream)
Curt! In order to think "I would like to eat ice cream", you would:
1.) Have to comprehend what "eating" is.
2.) Comprehend that you are not bound by hunger alone in order to eat.
3.) Comprehend what Ice Cream is.
4.) Comprehend that if you eat it again you will experience the same
pleasent sensations as the last time in which you ate it.
Now, let me ask you this. When is the last time you craved a food that
you have never eaten before?
>, where as comprehension means that we understand the
> implications of some data (I comprehend that the sign meant that I had to
> stop).
That's not all...You need to comprehend what a "sign" is, comprehend
the risks if you don't follow or obey the rule. You would have to
comprehend what a "rule" is period before being able to comprehend the
rules of the road!
>
> > > > 2.) I take it further. Your mind is simply the sum of that which you
> > > > comprehend. Accordingly, you can never think of anything other than
> > > > that wich you comprehend, even if only on the lowest of levels.
> > > > Don't believe me? Then think of something in which you possess
> > > > absolutely no comprehension of? What I am trying to get across is
> > > > that which we don't comprehend, is in itself, unconscious. So
> > > > therefor comprehension and conscious are the same exact thing.
> > >
> > > Ok, fine. But all you have said here is that we are consciously aware
> > > of what we are aware of and that we are not aware of the things we are
> > > not aware of. Isn't that all kinda obvious?
No. I want you to switch paradigms. In fact I will tell you what to
switch them to. I watch you to reread what I have said with the
assumption that I am supperior to you, and that the information is
correct. (I don't believe myself supperior to anyone, but people often
put emphasis on the words from people they consider such.) I also want
you to pretend that you have no prior opinions on the subject material.
> >
> > My comments above hopefully will clarify this further for you. You are
> > that which you understand!
>
> Sure, I agree with that. Or at least, my mind is that which I understand.
> My body is more than just my mind and there is much about my body which I
> do not understand but which I still call part of me.
O.K. thought I didn't bring up the body, I can now use it as an example
to try to reach you,
You posess self-image. The self image is based on part by:
1.) The information from sensory input gathered on your body.
2.) The extrapolation, of the sensory input information to form a
"whole". (You cannot see all of your body, You cannot feel all of the
body, etc....your brain has to make up this information or else you
never think about it!
3.) The Illusion of self, based on desires, opinions, and expectations.
4.) The illusion of self compared to the external environement
5.) The illusion of self compared to other people.
6.) The feedback of self from other people.
All of these go into forming your self image.
So, what you know of yourself is the information which you are both
conscious of and comprehend! The two can't be seperated.
Consciousness equals comprehension.
So what?
Well, there are flaws in your self image! You deped on feedback from
others to form your self-image. What if said people lied to you? What
if they just simply refused to inform you about a gross aspect of
yourself? It woud be a part of you, but it would not be part of your
self-image, consciousness, comprehension, or even mind. Using your own
words, "it would be a part of you", but you would never know it!
There are so many flaws in Turing Theory that you have simply
"accepted".
Simply "accepting" information implies that you don't consciously
understand
nor percieve the flaws. Try going over the above paragraphs an attempt
to find
the flaws. Apply logic and reason. Turing sure as hell didn't. He
made an
assumption, the assumption that he understood human intelligence.
Since
he believed he understood, he was more than thrilled to discover ways
in
which to test his "theory" of such. His theories only apply to this
his initial
assumption. Just as in the "self-image" example above, there is
information vital
to creating a complete picture of self-image that is unobtainable by
you. The
same thing applies for Turing. He was not capable of comprehending
that his
theories are flawed, merely becasue they are inadequatley incomplete.
>
> > The tests given for comprehension are given to children to determine
> > their ability to apply infromation gathered from disimilar experiences
> > and apply them toward solving a new problem. They are quite simple for
> > most adults to perform. It is not a Turing test as each answer is
> > depends on the creativity of the person taking the test. A turing test
> > could never erase prior programs, create new programs by combining
> > parts from other program, or disobey their programming.
>
> Your English is getting real odd here. "A Turing test could never erase
> programs"? What are you trying to say? You seem to be failing the Turing
> test. :)
replace "test" with "machine". I'm suprised this proved difficult for
you. Are you
sacrificing abilities to become more like Turing Machines as a way
around the problem?
If we make man just as unintelligent as Turing Machines then they will
share the same capacity.
>
> If you mean, "the computer being tested in the Turning test could never
> erase programs", you are just wrong. A computer program that changes it's
> program is called a learning machine. There are many different algorithms
> for creating learning machines - and the one thing they all have in common
> is that they change their programing when they learn - after they learn,
> they act differently than before they learned something.
(Yawn) They have to be programmed to be able to change their programing
before any such behavior results.
>
> It is true that a computer can never disobey it's programing (ignoring the
> issue of what happens when it "breaks"). But humans also are unable to
> disobey their "programing" which is defined by the laws of nature in
> combination with the physical structure of their body. A hydrogen atom can
> never choose to act like an oxygen atom. It's forced to follow it's
> programing and only act like a hydrogen atom. This is true for any and all
> parts of the human body.
Illogical. You are claiming that the laws of nature are programing.
That is a very low form of comprehension. Again you need to switch
paradigms. The laws of the universe are more properly refered to as
"Opperative Restraints" not "Programming". Operative Restraints are
needed to define the environement, the playing field, the stage, etc..
Operative Restrainst for humans would be DNA.
>
> There's nothing a human does that can not be explained as the human body
> just following its "programing". We can't easily prove that the human body
> is forced to follow the laws of nature but it's a fairly safe bet to assume
> it does.
There are countless things about the body wich we don't yet comprehend.
Teeth are one. We know teeth are alive because when they die they
turn black or gray. What we don't know is why they are able to remain
alive since an "adult" tooth does not recieve any blood supply. Only
developing teeth recieve blood supply. One theory suggests that the
tooth cells are able to obtain the food and oxygen they need from the
salivia, much like coral do in the ocean, but we have no proof of this.
>
> > Think about this one... A machine that does not possess intelligence
> > could not commit "suicide" unless programmed too.
>
> Think about this one. A human can not commit suicide unless the laws of
> nature (gravity, weak and strong atomic forces, etc), forced it to.
Again, you are using the wrong paradigm and as a result your
comprehension is low.
The laws of the univers are merely Operative Restraints, they are the
walls to bounce a ball off of, not any "programming".
>
> > Otherwise, an A.I.
> > with comprehension can freely choose to destroy itself and could
> > possibly leave a "suicide note" explaining it's decision and why it did
> > not wish to exist anymore.
>
> We already understand how to write computer programs that have the power to
> decide on their own, when they want to commit suicide. See Microsoft
> Windows and the blue screen of death. It even writes you a little suicide
> note and it decides all on it's own when it wants to do that.
Again, it must be pre-programmed for such behavior. Agian, it is not
conscious nor does it comprehend what it is doing. Nor does it
comprehend the words it displays.
>
> If you think a human committing suicide is doing something substantially
> different, then I would argue you have no clue what AI and human
> intelligence is all about. You are bestowing magic on us which just isn't
> there.
Or maybe you are telling me more about your own feeling of self than
you realize.
No. mimicking the cause of comprehension in humans is different from
mimicking intelligence. Mimicking the cause of human comprehension
will produce a similar comprehension in machines. The comprehension
will be genuine and severe, with machines making just as many errors
and faulty reasoning as you have done in your
replies. :)
"Blind leading the Blind" and "Room full of Monkey's typing.." come to
mind.
> But at the same time, it's highly likely that all previous AI work got some
> part of it correct. We just won't know which part they got correct and
> which they got wrong until someone finds the right answer.
>
> Everyone that's worth anything knows we need a new approach and they are
> already looking for it.
>
> What you have written however seems to indicate that your major approach is
> to use words like "comprehend" and "life" instead of words like "computer
> software" and "Turing tests". I don't see that as a new approach. I see
> that mostly as just ignoring the hard problem - which is how do you define
> "life" and "comprehension" in terms of wheels, and levers, and transistors.
You feel this way because you do not comprehend me, or your
comprehension
is extremely low. Trying pressing further.
> But maybe, you do have ideas about exactly how to do that and you just
> don't want to share it yet.
I can't share with a person unable/unwilling to comprehend.
-Jason
Immortalist
> >All humans talk but no house pets or house plants do, no matter how
>>pampered, so heredity must be involved in language. But a child growing up
>>in Japan speaks Japanese whereas the same child brought up in California
>>would speak English, so the environment is also crucial.
>
> Uh....Not in your analogy. Humans posess the capacity for Language
> period.
> It doesn't matter if it's Japanese, English, or any other spoken
> Language for that
> matter. All spoken languages use the medium of the vocal chords,
> throat, mouth,
> and tongue to create and control sounds. Therefore any human can learn
> any
> spoken language.
>
I agree but please point out where the text says that people can't learn
some languages. All human and animal languages are based upon the
subject/verb/object distinction or who did what to who or what did what to
what. Human beings have a particular set of noises that are juxtiposed over
this language instinct and have the bes ability to slow the now and control
memories.
>> Thus there is no
>>question about whether heredity or environment is involved in language, or
>>even whether one or the other is "more important."
>
> Incorrect. Yes we do know of cases where "environment" is vital to the
> development of language. If a small child is kept locked up, and not
> able to both hear language being used, then the child will never be
> able to properly grasp Language when older. This is because of a
> processes called "Neural Reinforcement", as well as "Plasticity" of the
> brain. If the neural pathways dedicated to spoken language are not
> stimulated within the "window of opportunity" allotted/provided to
> children, then the brain will divert the pathways and assign them
> different tasks.
>
Both are involved thats what I was saying, both are necessary; environmental
influences and instinctual language aquisition wiring.
Some very strong theories are about how second languages, learned after the
critical stages for language aquisition, involve other areas of the brain
that feed in and control Brocas and Wernickes areas whos activities produce
speech through controlling muscles.
I like to think of language accents. The accent you learn by the time you
reach puberty is the accent you have for life. You can learn other accents
after that by moving to other countries or areas but they are learned and
stored in other areas of the brain than the primary accent. Learning an
accent later in life is like creating a temporary work around that "must be
practiced or lost." Whereas the primary accent you developed around puberty
does not have to be practiced but can re-emerge in whole as a sort of
learned instinct.
Here is a list of the 200 or more instinctual drives in human nature, each
needing accenting at critical stages;
APPENDIX Donald E. Brown's List of Human Universals
THIS LIST, COMPILED in 1989 and published in 1991, consists primarily of
"surface" universals of behavior and overt language noted by ethnographers.
It does not list deeper universals of mental structure that are revealed by
theory and experiments. It also omits near-universals (traits that most, but
not all, cultures show) and conditional universals ("If a culture has trait
A, it always has trait B"). A list of items added since 1989 is provided at
the end. For discussion and references, see Brown's Human Universals (1991)
and his entry for "Human Universals" in The MIT Encyclopedia of the
Cognitive Sciences (Wilson & Keil, 1999).
abstraction in speech and thought
actions under self-control distinguished from those not under control
aesthetics
affection expressed and felt
age grades
age statuses
age terms
ambivalence
anthropomorphization
antonyms
baby talk
belief in supernatural/ religion
beliefs, false
beliefs about death
beliefs about disease
beliefs about fortune and misfortune
binary cognitive distinctions
biological mother and social mother normally the same person
black (color term)
body adornment
childbirth customs
childcare
childhood fears
childhood fear of loud noises
childhood fear of strangers
choice making (choosing alternatives)
classification
classification of age
classification of behavioral propensities
classification of body parts
classification of colors
classification of fauna
classification of flora
classification of inner states
classification of kin
classification of sex
classification of space
classification of tools
classification of weather
conditions coalitions
collective identities
conflict
conflict, consultation to deal with
conflict, means of dealing with
conflict, mediation of
conjectural reasoning
containers
continua (ordering as cognitive pattern)
contrasting marked and nonmarked sememes (meaningful elements in language)
cooking
cooperation
cooperative labor
copulation normally conducted in privacy
corporate (perpetual) statuses
coyness display
crying
cultural variability
culture
culture/nature distinction
customary greetings
daily routines
dance
death rituals
decision making
decision making, collective
directions, giving of
discrepancies between speech, thought, and action
dispersed groups
distinguishing right and wrong
diurnality
divination
division of labor
division of labor by age
division of labor by sex
dreams
dream interpretation
economic inequalities
economic inequalities, consciousness of
emotions
empathy
entification (treating patterns and relations as things)
environment, adjustments to
envy
envy, symbolic means of coping with
ethnocentrism
etiquette
explanation
face (word for)
facial communication
facial expression of anger
facial expression of contempt
facial expression of disgust
facial expression of fear
facial expression of happiness
facial expression of sadness
facial expression of surprise
facial expressions, masking/modifying of
family (or household)
father and mother, separate kin terms for
fears
fears, ability to overcome some
feasting
females do more direct childcare
figurative speech
fire
folklore
food preferences food sharing
future, attempts to predict
generosity admired
gestures
gift giving
good and bad distinguished
gossip
government
grammar
group living
groups that are not based on family
hairstyles
hand (word for)
healing the sick (or attempting to)
hospitality
hygienic care
identity, collective
incest between mother and son unthinkable or tabooed
incest, prevention or avoidance
in-group distinguished from out-group(s)
in-group, biases in favor of
inheritance rules
insulting
intention
interest in bioforms (living things or things that resemble them)
interpreting behavior
intertwining (e.g., weaving)
jokes
kin, close distinguished from distant
kin groups
kin terms translatable by basic relations of procreation
kinship statuses
language
language employed to manipulate others
language employed to misinform or mislead
language is translatable
language not a simple reflection of reality
language, prestige from proficient use of
law (rights and obligations)
law (rules of membership)
leaders
lever
linguistic redundancy
logical notions
logical notion of "and"
logical notion of "equivalent"
logical notion of "general/particular"
logical notion of "not"
logical notion of "opposite"
logical notion of "part/whole"
logical notion of "same"
magic
magic to increase life
magic to sustain life
magic to win love
male and female and adult and child seen as having different natures
males dominate public/political realm
males more aggressive
males more prone to lethal violence
males more prone to theft
manipulate social relations
marking at phonemic, syntactic, and lexical levels
marriage
materialism
meal times
meaning, most units of are non-universal
measuring
medicine
melody
memory
metaphor
metonym
mood- or consciousness-altering techniques and/or substances
morphemes
mother normally has consort during child-rearing years
mourning
murder proscribed
music
music, children's
music related in part to dance
music related in part to religious activity
music seen as art (a creation)
music, vocal
music, vocal, includes speech forms
musical redundancy
musical repetition
musical variation
myths
narrative
nomenclature (perhaps the same as classification)
nonbodily decorative art
normal distinguished from abnormal states
nouns
numerals (counting)
Oedipus complex
oligarchy (de facto)
one (numeral)
onomatopoeia
overestimating objectivity of thought
pain
past/present/future
person, concept of
personal names
phonemes
phonemes defined by sets of minimally contrasting features
phonemes, merging of
phonemes, range from 10 to 70 in number
phonemic change, inevitability of
phonemic change, rules of
phonemic system
planning
planning for future
play
play to perfect skills
poetry/rhetoric
poetic line, uniform length range
poetic lines characterized by repetition and variation
poetic lines demarcated by pauses
polysemy (one word has several related meanings)
possessive, intimate
possessive, loose
practice to improve skills
preference for own children and close kin (nepotism)
prestige inequalities
private inner life
promise
pronouns
pronouns, minimum two numbers
pronouns, minimum three persons
proper names
property
psychological defense mechanisms
rape
rape proscribed
reciprocal exchanges (of labor, goods, or services)
reciprocity, negative (revenge, retaliation)
reciprocity, positive
recognition of individuals by face
redress of wrongs
rhythm
right-handedness as population norm
rites of passage
rituals
role and personality seen in dynamic interrelationship (i.e., departures
from role can be explained in terms of individual personality)
sanctions
sanctions for crimes against the collectivity
sanctions include removal from the social unit
self distinguished from other
self as neither wholly passive nor wholly autonomous
self as subject and object
self is responsible
semantics
semantic category of affecting things and people
semantic category of dimension
semantic category of giving
semantic category of location
semantic category of motion
semantic category of speed
semantic category of other physical properties
semantic components
semantic components, generation
semantic components, sex
sememes, commonly used ones are short, infrequently used ones are longer
senses unified
sex (gender) terminology is fundamentally binary
sex statuses
sexual attraction
sexual attractiveness
sexual jealousy
sexual modesty
sexual regulation
sexual regulation includes incest prevention
sexuality as focus of interest
shelter
sickness and death seen as related
snakes, wariness around
social structure
socialization
socialization expected from senior kin
socialization includes toilet training
spear
special speech for special occasions
statuses and roles
statuses, ascribed and achieved
statuses distinguished from individuals
statuses on other than sex, age, or kinship bases
stop/nonstop contrasts (in speech sounds)
succession
sweets preferred
symbolism
symbolic speech
synonyms
taboos
tabooed foods
tabooed utterances
taxonomy
territoriality
time
time, cyclicity of
tools
tool dependency
tool making
tools for cutting
tools to make tools
tools patterned
culturally
tools, permanent
tools for pounding
trade
triangular awareness (assessing relationships among the self and two other
people)
true and false distinguished
turn-taking
two (numeral)
tying material (i.e., something like string)
units of time
verbs
violence, some forms of proscribed
visiting
vocalic/nonvocalic contrasts in phonemes
vowel contrasts
weaning
weapons
weather control (attempts to)
white (color term)
world view
---------------------
Additions Since 1989
---------------------
anticipation
attachment
critical learning periods
differential valuations
dominance/submission
fairness (equity), concept of
fear of death
habituation
hope
husband older than wife on average
imagery
institutions (organized co-activities)
intention
interpolation
judging others
likes and dislikes
making comparisons
males, on average, travel greater distances over lifetime
males engage in more coalitional violence
mental maps
mentalese
moral sentiments
moral sentiments, limited effective range of
precedence, concept of (that's how the leopard got its spots)
pretend play
pride
proverbs, sayings
proverbs, sayings-in mutually contradictory forms
resistance to abuse of power, to dominance
risk taking
self-control
self-image, awareness of (concern for what others think)
self-image, manipulation of
self-image, wanted to be positive
sex differences in spatial cognition and behavior
shame
stinginess, disapproval of
sucking wounds
synesthetic metaphors
thumb sucking
tickling
toys, playthings
...None of this means that people literally strive to replicate their genes.
If that's how the mind worked, men would line up outside the sperms banks
and women would pay to have their eggs harvested and given away to infertile
couples. It means only that inherited systems for learning, thinking, and
feeling have a design that would have led, on average, to enhanced survival
and reproduction in the environment in which our ancestors evolved. People
enjoy eating, and in a world without junk food, that led them to nourish
themselves, even if the nutritional content of the food never entered their
minds. People love sex and love children, and in a world without
contraception, that was enough for the genes to take care of themselves.
The difference between the mechanisms that impel organisms to behave in real
time and the mechanisms that shaped the design of the organism over
evolutionary time is important enough to merit some jargon. A proximate
cause of behavior is the mechanism that pushes behavior buttons in real
time, such as the hunger and lust that impel people to eat and have sex. An
ultimate cause is the adaptive rationale that led the proximate cause to
evolve, such as the need for nutrition and reproduction that gave us the
drives of hunger and lust. The distinction between proximate and ultimate
causation is indispensable in understanding ourselves because it determines
the answer to every question of the form "Why did that person act as he
did?" To take a simple example, ultimately people crave sex in order to
reproduce (because the ultimate cause of sex is reproduction), but
proximately they may do everything they can not to reproduce (because the
proximate cause of sex is pleasure).
The difference between proximate and ultimate goals is another kind of proof
that we are not blank slates. Whenever people strive for obvious rewards
like health and happiness, which make sense both proximately and ultimately,
one could plausibly suppose that the mind is equipped only with a desire to
be happy and healthy and a cause-and-effect calculus that helps them get
what they want. But people often have desires that subvert their proximate
well-being, desires that they cannot articulate and that they (and their
society) may try unsuccessfully to extirpate. They may covet their
neighbor's spouse, eat themselves into an early grave, explode over minor
slights, fail to love their stepchildren, rev up their bodies in response to
a stressor that they cannot fight or flee, exhaust themselves keeping up
with the Joneses or climbing the corporate ladder, and prefer a sexy and
dangerous partner to a plain but dependable one. These personally puzzling
drives have a transparent evolutionary rationale, and they suggest that the
mind is packed with cravings shaped by natural selection, not with a generic
desire for personal well-being.
Evolutionary psychology also explains why the slate is not blank. The mind
was forged in Darwinian competition, and an inert medium would have been
outperformed by rivals outfitted with high technology-with acute perceptual
systems, savvy problem-solvers, cunning strategists, and sensitive feedback
circuits. Worse still, if our minds were truly malleable they would be
easily manipulated by our rivals, who could mold or condition us into
serving their needs rather than our own. A malleable mind would quickly be
selected out.
Researchers in the human sciences have begun to flesh out the hypothesis
that the mind evolved with a universal complex design. Some anthropologists
have returned to an ethnographic record that used to trumpet differences
among cultures and have found an astonishingly detailed set of aptitudes and
tastes that all cultures have in common. This shared way of thinking,
feeling, and living makes us look like a single tribe, which the
anthropologist Donald Brown has called the Universal People, after Chomsky's
Universal Grammar. Hundreds of traits, from fear of snakes to logical
operators, from romantic love to humorous insults, from poetry to food
taboos, from exchange of goods to mourning the dead, can be found in every
society ever documented. It's not that every universal behavior directly
reflects a universal component of human natureâ?"many arise from an
interplay between universal properties of the mind, universal properties of
the body, and universal properties of the world. Nonetheless, the sheer
richness and detail in the rendering of the Universal People comes as a
shock to any intuition that the mind is a blank slate or that cultures can
vary without limit, and there is something on the list to refute almost any
theory growing out of those intuitions. Nothing can substitute for seeing
Brown's list in full; it is reproduced, with his permission, as an appendix.
The idea that natural selection has endowed humans with a universal complex
mind has received support from other quarters. Child psychologists no longer
believe that the world of an infant is a blooming, buzzing confusion,
because they have found signs of the basic categories of mind (such as those
for objects, people, and tools) in young babies. Archaeologists and
paleontologists have found that prehistoric humans were not brutish
troglodytes but exercised their minds with art, ritual, trade, violence,
cooperation, technology, and symbols. And primatologists have shown that our
hairy relatives are not like lab rats waiting to be conditioned but are
outfitted with many complex faculties that used to be considered uniquely
human, including concepts, a spatial sense, tool use, jealousy, parental
love, reciprocity, peacemaking, and differences between the sexes. With so
many mental abilities appearing in all human cultures, in children before
they have acquired culture, and in creatures that have little or no culture,
the mind no longer looks like a formless lump pounded into shape by culture.
FROM: The Blank Slate:
The Modern Denial of Human Nature
by Steven Pinker
http://www.amazon.com/exec/obidos/tg/detail/-/0142003344/
----------------------------------------
The ethnographic evidence overwhelmingly shows that a universal human nature
does exist. Brown lists nine pages of traits common to all known societies,
such as prestige, gossip, humorous insults, rhetoric, terms distinguishing
male and female, sexual regulations, kinship terms, property, rules
proscribing violence and rape, and many more.
This is hardly surprising. Human beings all have the same genomes. Genes
build bodies and bodies build brains and brains build minds. Ergo, human
beings are basically the same in the Amazon rainforests and the metropolitan
cities. Differences in behaviour, beliefs and habits - i.e. culture - are
mainly the result of ecology, geography and technology
The biology of culture
http://www.caribscape.com/baldeosingh/social/sober/2000/culture2.html
--------------------------------
The impression from anthropology that humanity is a carnival where anything
is possible came in part from a tourist mentality: when you come back from a
trip, you remember what was different about where you went, otherwise you
might as well have stayed at home. That is, many anthropologists exaggerated
the degree to which the tribes they studied were exotic and strange, both to
justify their profession and to raise people's consciousness about human
potential. But many of their claims have turned out either to be canards,
like Margaret Mead's claims about Samoa, or to miss the forest for the
trees: the anthropologists spent so much time looking for differences that
they didn't notice basic categories of human experience that are found in
every culture, like humor, love, jealousy, and a sense of responsibility.
Language is simply the most famous example of a human universal. Donald
Brown, an anthropologist at UC Santa Barbara, wrote a book called Human
Universals, in which he scoured the archives of ethnography for well
substantiated human universals. He came up with a list of about a hundred
and fifty, covering every sphere of human experience. That's my
interpretation of the main lessons of anthropology. The interesting
discoveries aren't about this kinship system or that form of shamanry.
Underneath it all - just as, in the case of language, there's a universal
design Chomsky called universal grammar - there is in the rest of culture
what Donald Brown calls the universal people. He characterized the human
species much the way a biologist would characterize any other species.
Chapter 13 - STEVEN PINKER
"Language Is a Human Instinct"
http://www.edge.org/documents/ThirdCulture/u-Ch.13.html
--------------------------------
Innateness
In linguistic and in non-linguistic domains, a central question is to what
degree "universals" that we observe are due to some innate property of human
brains, or to some kind of environmental influence -- the classic
nature/nurture problem.
Some things are obviously environmental. It's universally true that all
languages have a word for "mother" and "water", but this is because all
humans have these as part of their environment. But only some languages have
a word for "snow" or "alligator" because those are not universally present
features of the environment.
On the other hand, the general ability to learn language isn't
environmental -- otherwise any animal that grows up with humans would learn
to understand language (even if pronouncing it presented a problem). Recall
also that languages can be created spontaneously, as by deaf children with
limited sign systems who work together to make a fully functioning language
from these raw materials.
The important question is just how the universal language ability should be
characterized -- in essence, how detailed the instinct is. For example, are
we born with notions of hierarchical syntactic structure, or is this
something that we figure out based on more general notions of "kind" and
"superordinate"? Many researchers, inluding Pinker, believe that the
instinct is rather specific. Others look for more general explanations in
human cognition.
Analogies can be made to many other aspects of human behavior. Pinker
summarizes at length from Donald Brown's characterization of "universal
people", including things that are surely not innate, and others that likely
are.
The existence of shelter in every society is easily attributable to the
existence of dangers or inclement weather that can be mitigated by housing;
we probably don't need to propose a "shelter instinct" as different from a
general instinct for self-preservation.
The existence of facial expressions of various kinds is probably innate,
since they're the same across cultures, and in at least some cases seem to
be related to facial expressions in other primates (i.e. they've evolved
genetically).
The Standard Social Science Model (SSSM) assumes that human culture varies,
in principle, without limit.
http://www.ling.upenn.edu/courses/Spring_2003/ling001/14b.html
-----------------------------------
The most compelling attempt to date to build a thorough inventory of
humanity's basic toolbox comes from the anthropologist Donald Brown.
Inspired by Chomsky's idea of a "universal grammar"--the deep syntax shared
by all human languages--Brown set out to document the basic social patterns,
beliefs and categories shared by all known human societies, without
exception. Pinker devotes an entire appendix to Brown's list, which has a
strangely moving, abbreviated style: "cooking; cooperation; cooperative
labor; copulation normally conducted in privacy; corporate (perpetual)
statuses; coyness display; crying; cultural variability; culture;
culture/nature distinction; customary greetings; daily routines; dance;
death rituals..."
There is much for the left (and the right) to both condemn and admire in
this litany; for every "cooperative labor" there is a "females do more
direct childcare." But the first thing that should be noted about Brown's
list is its inclusiveness: We may in fact possess an innate tendency to
divide the world into "in groups" and "out groups," but the first instinct
of evolutionary psychology is to group us all together in the shared family
of human universals. Even if we don't always like the traits we find there,
that unifying impulse should be at the heart of any progressive politics,
not an outcast from it. Pinker quotes Chomsky on this very point:
A vision of a future social order is...based on a concept of human nature.
If, in fact, man is an indefinitely malleable, completely plastic being,
with no innate structures of mind and no intrinsic needs of a cultural or
social character, then he is a fit subject for the "shaping of behavior" by
the State authority, the corporate manager, the technocrat, or the central
committee. Those with some confidence in the human species will hope this is
not so and will try to determine the intrinsic characteristics that provide
the framework for intellectual development, the growth of moral
consciousness, cultural achievement, and participation in a free community.
Of course, the one place in which the neo-Darwinians have in fact emphasized
differences over commonalities is the fraught world of the sexes. Because so
much of natural selection is predicated on reproductive success or failure,
and because men and women have such differing biological stakes in the act
of reproduction, it is inevitable that natural selection would craft
slightly different toolboxes for each sex. This is no problem for the many
schools of feminism that embrace the "different but equal" assessment of the
sexes, but it is a major irritant for those on the left who imagine all
gender differences to be the product of cultural biases. I suspect, though,
that the sexual blank slate isn't long for this world, for several reasons.
Sociobiology and You
http://pinker.wjh.harvard.edu/slate_reviews_file/nation_johnson3.htm
----------------------------------------
Universals, Human Nature, and Anthropology
From HUMAN UNIVERSAL, Donald Brown (1991)
http://citd.scar.utoronto.ca/ANTB25/SCMEDIA/Readings/Brown.html
Anssi Hyytiainen
> The point about external behavior is that there's now way in hell you
> could get the external behavior correct, without getting the
> internal behavior correct. You can for example make a puppet look
> intelligent for about 5 seconds just by making it mimic intelligent
> actions.
Of course, I realize this. I'm just saying there are a lot of people out
there who are looking at the external behaviour and thinking about how
to mimic that, without putting any thought into what the internal
intelligence could be. Isn't that exactly what Jeff is going on about in
the beginning of his book?
It's not like Turing had the wrong bottom line, it's just that he didn't
seem to be concerned of trying to figure out any internal functions that
spring our external behaviour. Certainly a lot of people listening to
him didn't concern themselves with such things. Turing's test is very
bad idea in that it implies it is enough to mimic the external
behaviour. That it would somehow suffice if you can "fool" some people
into thinking that a corpse is still alive by pulling strings.
I am very tired of hearing such ideas as that more processing power is
"more intelligent". "When we get quantum computers then they will
surpass human brain in processing power". We both know how stupid that is.
> Everything we understand about intelligence in others is based on
> what we see happening with their external beahvior. If we saw _all_
> the same behaviors in a machine, would we be forced to conclude it
> was just as conscious as the humans which we see acting the same way.
Sure, but if I had programmed an algorithm that was mimicing behaviour,
I wouldn't consider passing Turing test to have any value at all. I
would know as a programmer what I've put in, and the machine would need
to consistently surprise me with behaviour I'd think is impossible from
what I had programmed in. It would have to create external behaviour and
personality spontaneously.
I'm not saying your approach couldn't do this. Your networks might just
as well produce just such a learning behaviour that I've described.
Still it would not be the learning behaviour that was conscious, it's
the abstraction level above that. And you would know for sure that real
intelligence had evolved because you'd know there's no single bit of
"intentionality" implemented by yourself into the system.
-Anssi
Anssi Hyytiainen
> Is Behavior the only thing we can study? Talk to me more about this.
> I already have some "deep thoughts" on this.
I'm not sure what you are asking about... Our semantical mind is
private, and at least I cannot think if any way to let two minds observe
each others directly. I can watch the electric impulses in your brain,
but they hold no such meaning to me that would tell me what you are
experiencing, or demonstrate that you are conscious. I don't know if you
are conscious. I can only assume you are. Maybe.
AFAIK it is technically possible to have two minds in one brain, but
that's just a case of two different "worldviews" being completely
separated from each others. The other might be religious, the other
might hate religion. One person/worldview wouldn't have access to the
ideas of the other. They wouldn't know how the other mind thinks and why.
They might switch places periodically, or there might be one worldview
in one hemisphere and one in the other so that they are both active all
the time (such is probably the case with Kim Peek). The communication
between them is basically external, just like that of two persons. They
probably even misunderstand each others' ideas just like two separate
brains, if they are made aware of them.
There are split personality cases that basically behave this way. I've
heard that the treatment of such cases is sometimes a therapy where
these two different personalities are made aware of each others as well
as possible, so that they will understand what is going on, and possibly
eventually merge the two worldviews into one. Of course society treats
such cases as "one person with a disorder", but basically if it was
beneficial to our survival, we would all be like that, and it would be
completely natural :)
Well, that's not proably what you were asking about, but hey...
-Anssi
Anssi Hyytiainen
> Language is a distinct piece of the biological makeup of the brain (18). The
> universality of complex language is the first reason to suspect that
> language is the product of a special human instinct. If language is an
> instinct, it should have an identifiable seat in the brain, and perhaps even
> a special set of genes that help wire it into place (45-46).
There is generally far too much importance placed on language when
people think about consciousness. You may think about many things by
thinking of words, but you also think of many things without any words.
Often times you know the idea you want to express, but don't have any
words to do it. And you can think about how a ball might bounce if you
throw it down the hall, without any language being part of it. And you
can think about pictures and diagrams, they are not language per se.
It should be revealing to do research on the practical problems of
implementing any human-like behaviour artificially. Like just walking,
or voice recognition, or pointing out boys and girls from a picture.
They are all similar semantical procedures as language, and it would be
madness to think that language would somehow be more important and even
have its own dedicated areas in brain.
Especially since there are many orthogonally different sorts of
instances of communication we think of as "language", and we are capable
of understanding all of them equally well. Spoken language is completely
different from written language. Former is sound patterns. Latter is
visual symbols. Deaf people don't know what language sounds like. Blind
people don't know what language looks like.
Then there is sign language. And we can even understand language by just
looking at someones lips when they talk. Or even by merely feeling the
talking lips (like Helen Keller). Or understand written language not by
seeing symbols but by feeling little dots on paper.
The power of intelligence is its rapid adaptability to its environment.
We are the adaptables. We overcome difficulties by wits.
> All humans talk but no house pets or house plants do, no matter how
> pampered, so heredity must be involved in language.
Gorillas can talk. They don't have the throat to produce words by voice,
but they can do sign language and understand spoken language. They can
understand abstract ideas and apply the words they have learned
intelligently. Koko asked her caretaker where will she go when she dies.
She was self-conscious. Koko was capable of learning enough to figure
out her own existence and also master language.
Language may help spread ideas between the members of a community, but I
don't think self-consciousness follows language. I think the gorillas in
nature are most likely also self-conscious, even if they are never
taught human language.
-Anssi
Anssi Hyytiainen
think about it. The structure is important. Chinese room exposes the
fatal flaw in such process structure that people often think of as
"intelligent". The flaw is to try to produce intelligent behaviour
directly through "raw" algorithm.
In the other posts I've discussed about such structure that forces an
algorithm to build "semantical database" in an on-going manner, without
having any explicit knowledge at its disposal. Intelligent behaviour
springs from that sort of structure. That is how human mind works. It is
naive to think human mind could explicitly know anything. Intelligent
design is a flawed idea in more ways than one :)
"Everything possible to be believed is an image of truth... Truth can
never be told so as to be understood, and not be believed."
-Anssi
Lucky
Anssi Hyytiainen wrote:
> Lucky wrote:
>
> > Is Behavior the only thing we can study? Talk to me more about this.
> > I already have some "deep thoughts" on this.
>
> I'm not sure what you are asking about... Our semantical mind is
> private, and at least I cannot think if any way to let two minds observe
> each others directly. I can watch the electric impulses in your brain,
> but they hold no such meaning to me that would tell me what you are
> experiencing, or demonstrate that you are conscious. I don't know if you
> are conscious. I can only assume you are. Maybe.
Is it possible to study one's self?
Think long and hard. Once you form an opinion, scrap it and think some
more!
>
> AFAIK it is technically possible to have two minds in one brain, but
> that's just a case of two different "worldviews" being completely
> separated from each others. The other might be religious, the other
> might hate religion. One person/worldview wouldn't have access to the
> ideas of the other. They wouldn't know how the other mind thinks and why.
Incorrect. If the mind is a by-product of the brain, then two "minds"
can just as easily co-exist. Hell, don't stop there...how about 4,5, or
6 of them?
More than one mind can exist in one brain as long as those minds never
become aware of each other. You have a "private mind", specificly
because of what I call "The Mind Barrier". If the barrier is breached
by another mind, then what I call "Mind Spillage" and "Mind Pooling"
will occur. "Mind Spillage" is a term to describe telepathy as a means
of communication. In the movies, Telepathy is portrayed as just
hearing a voice in your head. Unfortunately it wouldnt' work that way
in real life. If your mind is accessed via another mind, then the
minds will "spill" into each other, creating telepathy Information is
mutual spilled into the other mind by parties.. Now comes the
remarkable thing...."Mind Pooling". In "Mind Pooling", two seperate
minds merge to form one brand new third mind. Unlike with other things
that reproduce or come together, this third mind will show no
similarities to either parent mind. It will also be indistinguishable
between anyone else's mind. When seperation occcurs, and one mind is
withdrawn, the "Mind Pooling" will vanish. Individual, and private,
minds shall form again, though there might be an adjustment time where
an Identity Crisis might result. I suggest that the best solution to
this would be to jump into something new, gaining new experience, which
will help in the seperation process.
>
> They might switch places periodically, or there might be one worldview
> in one hemisphere and one in the other so that they are both active all
> the time (such is probably the case with Kim Peek). The communication
> between them is basically external, just like that of two persons. They
> probably even misunderstand each others' ideas just like two separate
> brains, if they are made aware of them.
>
> There are split personality cases that basically behave this way. I've
> heard that the treatment of such cases is sometimes a therapy where
> these two different personalities are made aware of each others as well
> as possible, so that they will understand what is going on, and possibly
> eventually merge the two worldviews into one. Of course society treats
> such cases as "one person with a disorder", but basically if it was
> beneficial to our survival, we would all be like that, and it would be
> completely natural :)
I don't believe in "split personality" or "Multiple Personality". I
believe these
individuals are indeed suffering from psychological disorder, but you
have to
understand that starting in the late 70's through today, every single
psychologist,
psyciatrist, or medical doctor has put forth a theory of a brand new
"disease" or
disorder. Why not just seperate those patients with schizophrenia that
have visions
of a giant mouse staring at them from those patients who believe
themselves to be mouses. We can then calim that the latter is caused
by Lycanthropy while the former
is a result of agoraphobia coupled with a brand new "disease" deemed,
"Jumbo Rodent Complex". We can justify our decisions by gathering
evidence that those poor souls who suffer from "Jumbo Rodent Complex",
have the same symptoms in common. Now,
we must then purge our minds or repress the fact that we formerly
viewed them as being schizophrenics as this could be used to disprove
our theory. We then diagnosis a few patients and then lay low. After
a time has passed, a suferer will come forward presenting this "new
disease" to the media, and hence the public. Both will assume that
since the diagnosis was made long ago, that this was just another
example of a rare disease and would accept it as a brand new disorder
without any further questions.
-Jason
Curt Welch
> The laws of the univers are merely Operative Restraints, they are the
> walls to bounce a ball off of, not any "programming".
So you are saying that a program not Operative Restraints for a computer?
Anthony G. Rubino
To Curt Welch:
The best example of a 'black box' AI device that we have is the
internet. I can't think of a better model, or resource. It already has
the largest data base we can devise and its growing as fast as any data
base can. It also has the most sophisticated search engines to find
relevant data that we have been able to devise. Everyone who presents a
question gets back answers that they believe has human origins. Pilot
projects to test AI algorithisms by posting to news groups like this
might prove interesting and lead to further development of stragedies
that could be symbiotic. On a lighter note, I have wondered at times
whether or not I was dealing with such a project.
Tony, philosopher
http://www.geocities.com/trisector/
So many misconceptions, so little time.
Lucky
>-Curt Welch
No. I am saying you lack the ability to comprehend the diference
between Programing and Operative Restraints.
-Jason
Curt Welch
> Curt Welch wrote:
> > The point about external behavior is that there's now way in hell you
> > could get the external behavior correct, without getting the
> > internal behavior correct. You can for example make a puppet look
> > intelligent for about 5 seconds just by making it mimic intelligent
> > actions.
>
> Of course, I realize this. I'm just saying there are a lot of people out
> there who are looking at the external behaviour and thinking about how
> to mimic that, without putting any thought into what the internal
> intelligence could be. Isn't that exactly what Jeff is going on about in
> the beginning of his book?
There are people having fun creating chat-bots and toy robots. I doubt any
of them really think they are creating intelligence. At best, they might
believe that are helping to advance the art so that one day we will be able
to create something intelligent.
> It's not like Turing had the wrong bottom line, it's just that he didn't
> seem to be concerned of trying to figure out any internal functions that
> spring our external behaviour.
Have you even read his paper? It doesn't sound like you have. Try reading
it and then say that with a straight face:
http://www.abelard.org/turpap/turpap.htm
He understood more about the problem 50 years ago than most people playing
with the idea of AI today. He very much was interested in understanding
how to implement intelligence.
> Certainly a lot of people listening to
> him didn't concern themselves with such things. Turing's test is very
> bad idea in that it implies it is enough to mimic the external
> behaviour. That it would somehow suffice if you can "fool" some people
> into thinking that a corpse is still alive by pulling strings.
Lots of people who talk about the "Turing test" have no clue what Turning
was talking about. You sound like one of them. Read his paper and stop
listening to all the fools that have no clue what Turning was talking
about.
> I am very tired of hearing such ideas as that more processing power is
> "more intelligent". "When we get quantum computers then they will
> surpass human brain in processing power". We both know how stupid that
> is.
Yeah, well, there are many fools out there.
> > Everything we understand about intelligence in others is based on
> > what we see happening with their external beahvior. If we saw _all_
> > the same behaviors in a machine, would we be forced to conclude it
> > was just as conscious as the humans which we see acting the same way.
>
> Sure, but if I had programmed an algorithm that was mimicing behaviour,
> I wouldn't consider passing Turing test to have any value at all.
You should. Have you ever tried to do it? Have you played the Turning
test with a program written to attempt to mimic intelligence? People have
tried real hard to make programs that simply won the Turning test by
fooling people. No one has come close because it can't be done. Either
you create real intelligence, or it will fail the Turning test.
Do you know there is a yearly contest with a $100,000 prize to the first
program that can create a machine to pass the Turing test? If mimicking
intelligence is so easy, why hasn't someone done it and picked up that
$100,000 prize?
Have you read the transcripts to see how "intelligent" the best programs
are these days? Try it:
http://loebner.net/Prizef/2005_Contest/Transcripts.html
They are complete jokes. You can't mimic intelligence. The only way for a
program to pass an extended Turing test, is to be intelligent.
> I
> would know as a programmer what I've put in, and the machine would need
> to consistently surprise me with behaviour I'd think is impossible from
> what I had programmed in. It would have to create external behaviour and
> personality spontaneously.
And if it didn't have the behavior you think it needed to be intelligent,
it would fail the Turning test - which is exactly what all attempts to date
have done. No program ever written has managed to fool anyone for more than
about 10 seconds. Most fail on the second question if not the first. Any
program that could last for 5 minutes in a Turning test would have to be
nearly as intelligent as an adult human.
> I'm not saying your approach couldn't do this. Your networks might just
> as well produce just such a learning behaviour that I've described.
.. and that Turning described 50 years ago. Read his paper.
> Still it would not be the learning behaviour that was conscious, it's
> the abstraction level above that. And you would know for sure that real
> intelligence had evolved because you'd know there's no single bit of
> "intentionality" implemented by yourself into the system.
I intentionally add strong learning function to the machine. Then I stop.
The learning hardware is what becomes the subconscious. Everything it is
conscious of, it would learn on it's own.
That's what I happen to believe is required for real intelligence.
Anssi Hyytiainen
> Have you even read his paper? It doesn't sound like you have. Try reading
> it and then say that with a straight face:
>
> http://www.abelard.org/turpap/turpap.htm
>
> He understood more about the problem 50 years ago than most people playing
> with the idea of AI today. He very much was interested in understanding
> how to implement intelligence.
The first indication of his lack of consideration comes here:
---
"The game may perhaps be criticised on the ground that the odds are
weighted too heavily against the machine. If the man were to try and
pretend to be the machine he would clearly make a very poor showing. He
would be given away at once by slowness and inaccuracy in arithmetic."
---
Obviously machine-like arithmetics are simple mechanics performed
without consciousess. That is orthogonally different from the method
with which whe perform arithmetics in our own head. We perform
arithmetics through our semantical mind, and that is why we are slow and
error prone.
As if the way to make a computer more human like was to add delay in the
responses to arithmetic problems, and perhaps even add a random error
here and there. In fact he proposes just such solution later on in the text:
---
"It is claimed that the interrogator could distinguish the machine from
the man simply by setting them a number of problems in arithmetic. The
machine would be unmasked because of its deadly accuracy. The reply to
this is simple. The machine (programmed for playing the game) would not
attempt to give the right answers to the arithmetic problems. It would
deliberately introduce mistakes in a manner calculated to confuse the
interrogator."
---
Delibarately introduce mistakes in a manner calculated to confuse the
interrogater?! Newsflash Mr. Turing, were you not just talking about
"imitation of external behaviour by an algorithm", you would realize
that true AI would be naturally just as slow and error-prone in
arithmethics as any real person.
And he keeps talking about intelligence in ways that clearly implies
that he thinks superficial mimicry of external behaviour equals to
intelligence:
---
"'If position 4505 contains 0 obey next the instruction stored in 6707,
otherwise continue straight on.'"
"Instructions of these types are very important because they make it
possible for a sequence of operations to be repeated over and over again
until some condition is fulfilled, but in doing so to obey, not fresh
instructions on each repetition, but the same ones over and over again.
To take a domestic analogy: suppose Mother wants Tommy to call at the
cobbler's every morning on his way to school to see if her shoes are
done, she can ask him afresh every morning. Alternatively she can stick
up a notice once and for all in the hall which he will see when he
leaves for school and which tells him to call for the shoes, and also to
destroy the notice when he comes back if he has the shoes with him."
"The reader must accept it as a fact that digital computers can be
constructed, and indeed have been constructed, according to the
principles we have described, and that they can in fact mimic the
actions of a human computer very closely."
---
They have been constructed and can mimic actions of human computer very
closely? Computer obeying orders is nothing like the process of a man
obeying orders.
And he says:
---
"Instead of trying to produce a programme to simulate the adult mind,
why not rather try to produce one which simulates the child's? If this
were then subjected to an appropriate course of education one would
obtain the adult brain. Presumably the child-brain is something like a
note-book as one buys it from the stationers. Rather little mechanism,
and lots of blank sheets. (Mechanism and writing are from our point of
view almost synonymous.) Our hope is that there is so little mechanism
in the child-brain that something like it can be easily programmed"
---
That he thinks it would be easier to program a child's brain than an
adult's brain implies that he thinks one must program the knowledge, the
rules of (conscious) conduct, into the system. That is precisely the
wrong idea. The algorithms that make the child's brain function are the
very same algorithms that make the adult's brain function, and whatever
knowledge these functions pick up from the environment is none of the
programmer's concern.
He goes on to discuss about experimentations where we would test how
well such an artificial mind would learn, which is a step to the right
direction, but there too he implies that the worldview was
pre-programmed to the state of that of a human child, and it would just
build on that. While it would be theoretically possible to program in a
worldview in a certain state, it really isn't something a programmer
should waste his time attempting to do himself. I doubt if even one's
memories could be reverse-engineered from the raw data into such form
that would make any sense to anyone else, let alone design consciously a
meaningful worldview from scratch.
He also mentions that the future states of a digital computer are much
more practically predictable than, for example, the future states of the
universe. But like we both know, if the digital computer is behaving as
subject to whatever it has and is learning in an open-ended fashion, you
cannot predict its future states any more than you can predict its
future experiences. It will "think" whatever the environment will make
it think.
And this goes on and on and on:
---
"It is probably wise to include a random element in a learning machine.
A random element is rather useful when we are searching for a solution
of some problem. Suppose for instance we wanted to find a number between
50 and 200 which was equal to the square of the sum of its digits, we
might start at 51 then try 52 and go on until we got a number that worked."
---
No, it is not wise for the programmer to include any random elements.
The programmer doesn't program any conscious experiences, such as
"guessing". Guessing behaviour comes out of the learning mechanisms
that the programmer includes.
Anyway, like I said, his bottom line argument is correct, our behaviour
is governed by explicit mechanism and is in principle predictable, just
like the behaviour of the universe is. But the problem is that he
implies at every turn that the programmer should concern himself with
coming up with such rules of conduct that would directly govern
intelligent-like external behaviour so to completely fool the judges at
the Turing's test. And that is precisely what most of AI community has
been doing since. I think you've seen this problem yourself and that's
why you don't concern yourself with attempting to program any rules that
would look like intelligent behaviour.
> Lots of people who talk about the "Turing test" have no clue what Turning
> was talking about. You sound like one of them. Read his paper and stop
> listening to all the fools that have no clue what Turning was talking
> about.
Obviously whatever Turing has said can be interpetated in many ways. If
I really want to, I can interpetate him sayings in such way that it
makes it appear like he has the exact same idea that I have. But I am
afraid if that is the case, he is not making himself very clear, and
then the problem is that great many people have gotten a completely
wrong idea from his sayings.
Anyway, I doubt this to be the case, because if Turing was indeed having
the same idea that I do, he would not have proposed such a thing as
Turing's test at all. When we build the first intelligent machine, we
will not need a Turing's test to find out. In fact, it is madness to
think that the first intelligent machine will even communicate in
english or any human-like language. It will be more like a cat or a dog
or an ape at most. ("Intelligence" here is the component pre-requisite
of consciousness) We will know because it it learning and applying
whatever it has learned into new situations in such a creative manner
that was not programmed in. None of its external behaviour is not
programmed in.
>>Sure, but if I had programmed an algorithm that was mimicing behaviour,
>>I wouldn't consider passing Turing test to have any value at all.
>
> You should. Have you ever tried to do it? Have you played the Turning
> test with a program written to attempt to mimic intelligence? People have
> tried real hard to make programs that simply won the Turning test by
> fooling people. No one has come close because it can't be done. Either
> you create real intelligence, or it will fail the Turning test.
I mean I wouldn't consider it to have any value at all in implication
whether or not the system has consciousness.
> Do you know there is a yearly contest with a $100,000 prize to the first
> program that can create a machine to pass the Turing test? If mimicking
> intelligence is so easy, why hasn't someone done it and picked up that
> $100,000 prize?
I'm not saying it's easy, I'm saying it's impossible. I'm saying it is
the wrong method in building intelligence. The attempts to win that
prize have nothing to do with trying to create conscious intelligence.
They are just technical masturbation.
> Have you read the transcripts to see how "intelligent" the best programs
> are these days? Try it:
>
> http://loebner.net/Prizef/2005_Contest/Transcripts.html
>
> They are complete jokes. You can't mimic intelligence. The only way for a
> program to pass an extended Turing test, is to be intelligent.
Yeah sure, and another joke is that sometimes the judges mistake real
persons to be machines. I don't know why people even bother with that test.
I think it would be interesting if people tried to crack some simpler
"semantical" task first. Like just a machine learning to move around in
an environment fluently, or something else that simple biological
organisms are capable of.
For example, it would be interesting apply your networks so to control a
physical body in a virtual environemnt. Just have some sort of thing
with senses and limbs, and allow it to move its limbs freely. And then
put some evolutionary pressure for it to learn to move around. To learn
to jump and to climb and whatever is necessary in any environment.
Perhaps even stack boxes.
Basically try to generate pressure for such a behaviour to arise that
would "understand" in some way the meaning of the motion of the limbs,
and how the world where it lives in works. And how to apply the motion
of limbs to reach places. Or whatever is declared to be "good" by the
critic (getting to "food", etc...)
-Anssi
Anssi Hyytiainen
> Anssi Hyytiainen wrote:
>
>>Lucky wrote:
>>
>> > Is Behavior the only thing we can study? Talk to me more about this.
>> > I already have some "deep thoughts" on this.
>>
>>I'm not sure what you are asking about... Our semantical mind is
>>private, and at least I cannot think if any way to let two minds observe
>>each others directly. I can watch the electric impulses in your brain,
>>but they hold no such meaning to me that would tell me what you are
>>experiencing, or demonstrate that you are conscious. I don't know if you
>>are conscious. I can only assume you are. Maybe.
>
>
> Is it possible to study one's self?
>
> Think long and hard. Once you form an opinion, scrap it and think some
> more!
It depends on what you mean by that question exactly. In principle there
must always be something in our mind that we cannot "comprehend", to use
your terminology. There can be no machine that is so complex that it
even understands its own complexity, since this understanding is part of
the complexity itself. The highest abstraction level of our mind can
only understand the lower level functions underneath itself, but not
itself, so to speak.
But that is not to say we couldn't build conscious machine. All our
ideas are just ideas. If one comes up with the right idea, we got it. We
would still not completely understand why all the mechanical activity
that is going on in this artificial mind makes it conscious though.
Anyway, that's the opinion I've had the longest. At first I even thought
it meant that we could never build a conscious machine, but I don't
think so anymore. The truth with it still prevails in that there are
elements in our conscious experience that cannot be "comprehended",
although it might at times seem that we already know "everything" there
is to know about it.
I can't really think of other options right now. Can you?
>>AFAIK it is technically possible to have two minds in one brain, but
>>that's just a case of two different "worldviews" being completely
>>separated from each others. The other might be religious, the other
>>might hate religion. One person/worldview wouldn't have access to the
>>ideas of the other. They wouldn't know how the other mind thinks and why.
>
> Incorrect. If the mind is a by-product of the brain, then two "minds"
> can just as easily co-exist.
Isn't that what I said? :) "It is possible to have two minds in one
brain." Read the posts :)
One mind not having an access to the other mind is just a definition of
what two minds means, which you repeat yourself:
> More than one mind can exist in one brain as long as those minds never
> become aware of each other.
-Anssi
Lester Zick
comp.ai.philosophy wrote:
>
>Lester Zick wrote:
>
>> Below you suggest the instruction language could be just a set of
>> pictures and arrows. But the question is really could it be? Let's
>> suppose we had such a translation grid. The problem then becomes not
>> so much whether it could produce answers but whether it can produce
>> correct answers. And what you'll find is that the kind of answers you
>> can get and the kind of significance which can attach to the symbols
>> depends exactly on the instruction language.
>
> You are not understanding that the "The Chinese Room" Arguement is
>flawed
Yes especially since I point out exactly that the argument is flawed
but that Anssi's analysis of the flaw is also flawed.
>for one thing, and one thing only.
Yes but long winded speeches and interminable dialectical lectures on
mechanics don't have anything to do with the flaw.
>I explained that when first attempting to understand the arguement we
>must first
>discover if any faulty logic was used in it's creation.
>
>Unfortunatley "The Chines Room" Arguement does contain faulty logic.
>
>It depends on system created of individual units, in this case people,
>who
>themselves possess Intelligence.
>
>Therefore, human intelligence can be utililysed to overcome any
>obstacle implied in the arguement itself. A human being can therefore
>follow the instructions and correctly produce the corresponding Chinese
>Character without any prior knowledge of the Chinese Language.
How exactly? All I see are claims. I don't see any proof. Anssi's
claim seems to be mainly that the puzzle is wrong because it suggests
intelligence is mechanical whereas my claim is not that it's wrong
because we're mechanical but that the puzzle uses the wrong mechanics.
>Going further, since Humans posess the ability to learn by repetition,
>experience/involvement, and context, the man in the room might be able
>to form a "self-taught" form of Chinese that in itself would pass as if
>he truly learned the language in a more "normal" fashion!
So you say. I don't think the problem is that people can learn Chinese
because billions have. The problem lies with the supposition there can
be a dictionary which produces uniformly correct answers to questions
in Chinese without someone thinking in Chinese to provide the answer.
>> Previously you complained symbols are only symbols and cognition is
>> more than mere symbols. And that's true. However what makes that true
>> is the kind of instruction language for translation and manipulation.
>> Hell even I could give a transliteration in Chinese but that wouldn't
>> make the answers correct to a native Chinese speaker. Nor would it
>> satisfy turing criteria for intelligence.
>
>Yes, but the reasons why this is so is due to an inherent flaw in human
>development. In order to properly understand chinese, all prior
>knowledge of any other language would need to be compartmentalized or
>somehow destroyed. The individual would then have to devote the same
>number of years, from infancy forward, to learning Chinese. If any
>other method is used, the brain has no choice but to use the primary
>language as a paradigm to mold the understanding of the secondary
>language.
Yeah sure. All kinds of "fundamental flaws" in our nature prevent
your claims from being true. Not the first time I've heard a universal
conspiracy theory used to excuse a poor excuse of a mechanical theory.
Curt's variation on the same theme is "we done it to ourselves with
language".
>Going further, The brain has the ability to swap primary and secondary
>languages. For example, if you took classes in French and then moved
>to France permanently, you would for a time still think in English and
>then translate that into French before speaking. However, in time this
>will change. From what I have learned from communicating with such
>people, is that their dreams begin to change. They start dreaming in
>the other language! Once they begin dreaming in the other language
>they then notice they are thinking in the other language. Therefore,
>French which was once the secondary language has now been switched to
>become the primary language!
>
>Note: I previously theorized that Dreams play a vital role in the
>process of comprehension/understanding.
Yeah look. So far all you've done is state a lot of explicit claims,
failed to back any of them up with mechanical arguments, tried to
support them with further unsupported claims, then tried to take
credit for resolving issues in AI. This is the essence of analogical
dialectical argumentation in philosophy where claims are supported
with other unsupported claims. It has nothing to do with mechanics or
science.
~v~~
1Z
Anthony G. Rubino wrote:
> To Curt Welch:
>
> The best example of a 'black box' AI device that we have is the
> internet. I can't think of a better model, or resource. It already has
> the largest data base we can devise and its growing as fast as any data
> base can.
OTOH, it doesn't have a lot of autonomy. More of a cyborg than a
AI.
Lucky
Yes. I will ask you again. Can you comprehend consciousness. Answer
with an initial "Yes" or "No", followed by a simple explanation.
My theory suggests that comprehension is a process. I believe we have
no choice but to comprehend information that has gone through such
process, even if it is entirely wrong. Have you noticed how we always
comprehend something, even if it is completely wrong!
So, can we comprehend consciousness? Yes, it's just that our
comprehension will be completely wrong since it is not based on
external sensory input. The comprehension that will arise is simply
self-deception. "The Deceptive Consciousness". So my next question
would be, are we capable of "seeing" beyond the comprehension process?
>
> >>AFAIK it is technically possible to have two minds in one brain, but
> >>that's just a case of two different "worldviews" being completely
> >>separated from each others. The other might be religious, the other
> >>might hate religion. One person/worldview wouldn't have access to the
> >>ideas of the other. They wouldn't know how the other mind thinks and why.
> >
> > Incorrect. If the mind is a by-product of the brain, then two "minds"
> > can just as easily co-exist.
>
> Isn't that what I said? :) "It is possible to have two minds in one
> brain." Read the posts :)
You did, sorry! :) Oh wait! Could our two minds be inhabiting the
same body?
AAAAHHHHHHHHHHHHHHHH!!!!
-Jason
Lucky
Lester Zick wrote:
> On 15 Feb 2006 11:25:45 -0800, "Lucky" <Lucky...@aol.com> in
> comp.ai.philosophy wrote:
>
> >
> >Lester Zick wrote:
> >
> >> Below you suggest the instruction language could be just a set of
> >> pictures and arrows. But the question is really could it be? Let's
> >> suppose we had such a translation grid. The problem then becomes not
> >> so much whether it could produce answers but whether it can produce
> >> correct answers. And what you'll find is that the kind of answers you
> >> can get and the kind of significance which can attach to the symbols
> >> depends exactly on the instruction language.
> >
> > You are not understanding that the "The Chinese Room" Arguement is
> >flawed
>
> Yes especially since I point out exactly that the argument is flawed
> but that Anssi's analysis of the flaw is also flawed.
>
> >for one thing, and one thing only.
>
> Yes but long winded speeches and interminable dialectical lectures on
> mechanics don't have anything to do with the flaw.
Quit unconsciously projecting qualities of yourself.
> >I explained that when first attempting to understand the arguement we
> >must first
> >discover if any faulty logic was used in it's creation.
> >
> >Unfortunatley "The Chines Room" Arguement does contain faulty logic.
> >
> >It depends on system created of individual units, in this case people,
> >who
> >themselves possess Intelligence.
> >
> >Therefore, human intelligence can be utililysed to overcome any
> >obstacle implied in the arguement itself. A human being can therefore
> >follow the instructions and correctly produce the corresponding Chinese
> >Character without any prior knowledge of the Chinese Language.
>
> How exactly? All I see are claims. I don't see any proof. Anssi's
> claim seems to be mainly that the puzzle is wrong because it suggests
> intelligence is mechanical whereas my claim is not that it's wrong
> because we're mechanical but that the puzzle uses the wrong mechanics.
I don't have to provide proof for this. You realize that "Chinese" is
a human
invention? A human being will behave in the manner of a human being.
He will
make many mistakes at first, as knowledge is gained he will make fewer
and fewer.
He will never reach perfection, but that's o.k. since no machine will
either.
>
> >Going further, since Humans posess the ability to learn by repetition,
> >experience/involvement, and context, the man in the room might be able
> >to form a "self-taught" form of Chinese that in itself would pass as if
> >he truly learned the language in a more "normal" fashion!
>
> So you say. I don't think the problem is that people can learn Chinese
> because billions have. The problem lies with the supposition there can
> be a dictionary which produces uniformly correct answers to questions
> in Chinese without someone thinking in Chinese to provide the answer.
Are you still caught up in your belief about there being a "universal
language"
involved in this arugement? If so, then yoiu need to read the thread
"Observation
and Assumptions in the Chinese Room Arguement", where I use logic to
disprove
such notion. The system is using the pre-existing abilities of the
individual componets in order to communicate. No spoken language is
used.
>
> >> Previously you complained symbols are only symbols and cognition is
> >> more than mere symbols. And that's true. However what makes that true
> >> is the kind of instruction language for translation and manipulation.
> >> Hell even I could give a transliteration in Chinese but that wouldn't
> >> make the answers correct to a native Chinese speaker. Nor would it
> >> satisfy turing criteria for intelligence.
> >
> >Yes, but the reasons why this is so is due to an inherent flaw in human
> >development. In order to properly understand chinese, all prior
> >knowledge of any other language would need to be compartmentalized or
> >somehow destroyed. The individual would then have to devote the same
> >number of years, from infancy forward, to learning Chinese. If any
> >other method is used, the brain has no choice but to use the primary
> >language as a paradigm to mold the understanding of the secondary
> >language.
>
> Yeah sure. All kinds of "fundamental flaws" in our nature prevent
> your claims from being true. Not the first time I've heard a universal
> conspiracy theory used to excuse a poor excuse of a mechanical theory.
> Curt's variation on the same theme is "we done it to ourselves with
> language".
Universal Conspiracy theory? What the hell are you talking about?
Look,
I don't feel you are correctly comprehending the information
communicated to
you. It's most likely an ego thing. Surrender your ego and you should
have no
further difficulties.
>
> >Going further, The brain has the ability to swap primary and secondary
> >languages. For example, if you took classes in French and then moved
> >to France permanently, you would for a time still think in English and
> >then translate that into French before speaking. However, in time this
> >will change. From what I have learned from communicating with such
> >people, is that their dreams begin to change. They start dreaming in
> >the other language! Once they begin dreaming in the other language
> >they then notice they are thinking in the other language. Therefore,
> >French which was once the secondary language has now been switched to
> >become the primary language!
> >
> >Note: I previously theorized that Dreams play a vital role in the
> >process of comprehension/understanding.
>
> Yeah look. So far all you've done is state a lot of explicit claims,
> failed to back any of them up with mechanical arguments, tried to
> support them with further unsupported claims, then tried to take
> credit for resolving issues in AI. This is the essence of analogical
> dialectical argumentation in philosophy where claims are supported
> with other unsupported claims. It has nothing to do with mechanics or
> science.
>
> ~v~~
You are completely wrong. The "mechanical" arguments are right there
under
your nose but you are too full of your self to be able to see them.
Drop the ego,
switch your paradigm, and comprehend the situation.
-Jason
Lester Zick
comp.ai.philosophy wrote:
>
>Lester Zick wrote:
>> On 15 Feb 2006 11:25:45 -0800, "Lucky" <Lucky...@aol.com> in
>> comp.ai.philosophy wrote:
>>
>> >
>> >Lester Zick wrote:
>> >
>> >> Below you suggest the instruction language could be just a set of
>> >> pictures and arrows. But the question is really could it be? Let's
>> >> suppose we had such a translation grid. The problem then becomes not
>> >> so much whether it could produce answers but whether it can produce
>> >> correct answers. And what you'll find is that the kind of answers you
>> >> can get and the kind of significance which can attach to the symbols
>> >> depends exactly on the instruction language.
>> >
>> > You are not understanding that the "The Chinese Room" Arguement is
>> >flawed
>>
>> Yes especially since I point out exactly that the argument is flawed
>> but that Anssi's analysis of the flaw is also flawed.
>>
>> >for one thing, and one thing only.
>>
>> Yes but long winded speeches and interminable dialectical lectures on
>> mechanics don't have anything to do with the flaw.
>
>Quit unconsciously projecting qualities of yourself.
My speeches are longer than yours? Just an unlucky guess, I guess.
>> >I explained that when first attempting to understand the arguement we
>> >must first
>> >discover if any faulty logic was used in it's creation.
>> >
>> >Unfortunatley "The Chines Room" Arguement does contain faulty logic.
>> >
>> >It depends on system created of individual units, in this case people,
>> >who
>> >themselves possess Intelligence.
>> >
>> >Therefore, human intelligence can be utililysed to overcome any
>> >obstacle implied in the arguement itself. A human being can therefore
>> >follow the instructions and correctly produce the corresponding Chinese
>> >Character without any prior knowledge of the Chinese Language.
>>
>> How exactly? All I see are claims. I don't see any proof. Anssi's
>> claim seems to be mainly that the puzzle is wrong because it suggests
>> intelligence is mechanical whereas my claim is not that it's wrong
>> because we're mechanical but that the puzzle uses the wrong mechanics.
>
>I don't have to provide proof for this.
Of course you don't. You provide plenty of claims but no proof. I
expect we'll just have to take your word for it. Seems we just have to
take your word for pretty much everything you say.
> You realize that "Chinese" is
>a human invention?
Not really.
> A human being will behave in the manner of a human being.
Yeah that's real bright.
>He will make many mistakes at first, as knowledge is gained he will make fewer
>and fewer. He will never reach perfection, but that's o.k. since no machine will
>either.
I think we can safely conclude you will indeed never reach perfection.
>> >Going further, since Humans posess the ability to learn by repetition,
>> >experience/involvement, and context, the man in the room might be able
>> >to form a "self-taught" form of Chinese that in itself would pass as if
>> >he truly learned the language in a more "normal" fashion!
>>
>> So you say. I don't think the problem is that people can learn Chinese
>> because billions have. The problem lies with the supposition there can
>> be a dictionary which produces uniformly correct answers to questions
>> in Chinese without someone thinking in Chinese to provide the answer.
>
>Are you still caught up in your belief about there being a "universal language"
>involved in this arugement? If so, then yoiu need to read the thread "Observation
>and Assumptions in the Chinese Room Arguement", where I use logic todisprove
>such notion. The system is using the pre-existing abilities of the
>individual componets in order to communicate. No spoken language is used.
My, my, seems you're better at disproving things than proving them.
>> >> Previously you complained symbols are only symbols and cognition is
>> >> more than mere symbols. And that's true. However what makes that true
>> >> is the kind of instruction language for translation and manipulation.
>> >> Hell even I could give a transliteration in Chinese but that wouldn't
>> >> make the answers correct to a native Chinese speaker. Nor would it
>> >> satisfy turing criteria for intelligence.
>> >
>> >Yes, but the reasons why this is so is due to an inherent flaw in human
>> >development. In order to properly understand chinese, all prior
>> >knowledge of any other language would need to be compartmentalized or
>> >somehow destroyed. The individual would then have to devote the same
>> >number of years, from infancy forward, to learning Chinese. If any
>> >other method is used, the brain has no choice but to use the primary
>> >language as a paradigm to mold the understanding of the secondary
>> >language.
>>
>> Yeah sure. All kinds of "fundamental flaws" in our nature prevent
>> your claims from being true. Not the first time I've heard a universal
>> conspiracy theory used to excuse a poor excuse of a mechanical theory.
>> Curt's variation on the same theme is "we done it to ourselves with
>> language".
>
>Universal Conspiracy theory? What the hell are you talking about? Look,
>I don't feel you are correctly comprehending the information communicated to
>you. It's most likely an ego thing. Surrender your ego and you should have no
>further difficulties.
Well Curt has his linguistic universal conspiracy theory and Glen his
cognitive universal conspiracy theory. Now it seems we have your own
variation with fundamental flaws to explain why your theories are true
but no one especially cares to believe what you say.
>> >Going further, The brain has the ability to swap primary and secondary
>> >languages. For example, if you took classes in French and then moved
>> >to France permanently, you would for a time still think in English and
>> >then translate that into French before speaking. However, in time this
>> >will change. From what I have learned from communicating with such
>> >people, is that their dreams begin to change. They start dreaming in
>> >the other language! Once they begin dreaming in the other language
>> >they then notice they are thinking in the other language. Therefore,
>> >French which was once the secondary language has now been switched to
>> >become the primary language!
>> >
>> >Note: I previously theorized that Dreams play a vital role in the
>> >process of comprehension/understanding.
>>
>> Yeah look. So far all you've done is state a lot of explicit claims,
>> failed to back any of them up with mechanical arguments, tried to
>> support them with further unsupported claims, then tried to take
>> credit for resolving issues in AI. This is the essence of analogical
>> dialectical argumentation in philosophy where claims are supported
>> with other unsupported claims. It has nothing to do with mechanics or
>> science.
>>
>> ~v~~
>
>
>You are completely wrong. The "mechanical" arguments are right there under
>your nose but you are too full of your self to be able to see them. Drop the ego,
>switch your paradigm, and comprehend the situation.
I much prefer to judge your words with my ego since you've just
explained to us that you won't prove what you say yet expect us to
believe what you say anyway.
~v~~
Allan C Cybulskie
> Going further, The brain has the ability to swap primary and secondary
> languages. For example, if you took classes in French and then moved
> to France permanently, you would for a time still think in English and
> then translate that into French before speaking. However, in time this
> will change. From what I have learned from communicating with such
> people, is that their dreams begin to change. They start dreaming in
> the other language! Once they begin dreaming in the other language
> they then notice they are thinking in the other language. Therefore,
> French which was once the secondary language has now been switched to
> become the primary language!
I once asked a co-worker of mine who was born in Macau what language she
thought in. She said that when she was at work she thought in English and
when she was around Chinese-speaking people outside of work she thought in
Chinese. Which was her primary language, under your scheme?
It seems to me that language is context/use sensitive, and not as sharply
defined as you claim it is.
Curt Welch
> Curt Welch wrote:
> > Have you even read his paper? It doesn't sound like you have. Try
> > reading it and then say that with a straight face:
> >
> > http://www.abelard.org/turpap/turpap.htm
> >
> > He understood more about the problem 50 years ago than most people
> > playing with the idea of AI today. He very much was interested in
> > understanding how to implement intelligence.
>
> The first indication of his lack of consideration comes here:
> ---
> "The game may perhaps be criticised on the ground that the odds are
> weighted too heavily against the machine. If the man were to try and
> pretend to be the machine he would clearly make a very poor showing. He
> would be given away at once by slowness and inaccuracy in arithmetic."
> ---
>
> Obviously machine-like arithmetics are simple mechanics performed
> without consciousess. That is orthogonally different from the method
> with which whe perform arithmetics in our own head. We perform
> arithmetics through our semantical mind, and that is why we are slow and
> error prone.
Yes, I agree. If we create intelligence in a computer, it would do math
the same way we do and it would make the type of errors we make. It
wouldn't be using it's internal math hardware for doing math. But you are
criticizing someone for not understanding intelligence in mechanical forms
as well as we do when we have 56 years of AI research and 56 years of
advances in computers behind us.
Turning suggested two ways to try and make a computer intelligent. The
first was to hard-code intelligent behavior into the machine. If you did
that, then you could and probably would, hard code math behavior using math
hardware and you would have the problem he suggested above - the computer
would be "too intelligent" when compared to a human in those areas, so you
would have to make it stupid in order to not give itself away.
The other way he suggested was to build a learning machine. In that case,
you would end up with a machine that did math like we did because it was
all just generic learned behavior which like all our behavior, is never
perfect.
> "The reader must accept it as a fact that digital computers can be
> constructed, and indeed have been constructed, according to the
> principles we have described, and that they can in fact mimic the
> actions of a human computer very closely."
> ---
>
> They have been constructed and can mimic actions of human computer very
> closely? Computer obeying orders is nothing like the process of a man
> obeying orders.
I don't think you have any grasp of history. Do you understand that most
people in those days had no clue what a computer was or how it worked?
there were only a small handful of computers (5?) in existence when he
wrote that. Mechanical adding machines were all most people knew about in
those days.
And do you have any clue what he was referring to when he wrote "human
computer"? He was talking about a guy sitting down with pencil, paper, and
adding machine, calculating sine, cosine or log tables by following
detailed written algorithms for doing those calculations.
The quote of his from above was just trying to explain to people what a
computer was. It had nothing to do with AI other than background to help
people that didn't know about computers to understand what they can do.
> And he says:
> ---
> "Instead of trying to produce a programme to simulate the adult mind,
> why not rather try to produce one which simulates the child's? If this
> were then subjected to an appropriate course of education one would
> obtain the adult brain. Presumably the child-brain is something like a
> note-book as one buys it from the stationers. Rather little mechanism,
> and lots of blank sheets. (Mechanism and writing are from our point of
> view almost synonymous.) Our hope is that there is so little mechanism
> in the child-brain that something like it can be easily programmed"
> ---
>
> That he thinks it would be easier to program a child's brain than an
> adult's brain implies that he thinks one must program the knowledge, the
> rules of (conscious) conduct, into the system. That is precisely the
> wrong idea. The algorithms that make the child's brain function are the
> very same algorithms that make the adult's brain function, and whatever
> knowledge these functions pick up from the environment is none of the
> programmer's concern.
You have no clue what he's talking about do you.
Do you think you could program a learning machine without writing any code
or building any hardware? Don't you think that whatever code you end up
writing, or hardware you end up building is in fact the "mechanism of the
child-brain"?
It's true that 56 years ago, he didn't understand as much as we do now.
But what is irritating me is that you are saying all these negative things
about Turning, and what he suggested we do 56 years ago, then you turn
around and suggest that what we should do instead, but you are too naive to
even understand that what you are suggesting, is exactly what he suggested
in that paper 56 years ago. If you can't see that, then you don't really
understand what in fact you are suggesting we do, which means you still
don't have the understanding he had 56 years ago.
> He goes on to discuss about experimentations where we would test how
> well such an artificial mind would learn, which is a step to the right
> direction, but there too he implies that the worldview was
> pre-programmed to the state of that of a human child,
Have you seen a new-born? That don't have a world view. They suck and cry
by hard-wired instinct and they have no clue that the hand in front of
their own face is their own. When Turning was talking "child mind" he
wasn't talking the mind of a 8 year old. He was talking the mind if child
who had not yet learned anything. That was his entire point - to minimize
what had to be build into the machine (only learning power and no
knowledge).
> And this goes on and on and on:
> ---
> "It is probably wise to include a random element in a learning machine.
> A random element is rather useful when we are searching for a solution
> of some problem. Suppose for instance we wanted to find a number between
> 50 and 200 which was equal to the square of the sum of its digits, we
> might start at 51 then try 52 and go on until we got a number that
> worked." ---
>
> No, it is not wise for the programmer to include any random elements.
> The programmer doesn't program any conscious experiences, such as
> "guessing". Guessing behaviour comes out of the learning mechanisms
> that the programmer includes.
Have you ever tried to build a learning machine? There's plenty of reason
to include some form of random element. If you learn by experience, and
and you always pick the behavior you believe to be the best, you will never
find out if some other behavior might be better because you will never get
the opportunity to test it. There's plenty of reason to build randomness
in at the lowest level of the learning function. However, for human
behavior, in real world environments, I believe the environment is complex
enough as to create all the randomness you need.
> Anyway, like I said, his bottom line argument is correct, our behaviour
> is governed by explicit mechanism and is in principle predictable, just
> like the behaviour of the universe is. But the problem is that he
> implies at every turn that the programmer should concern himself with
> coming up with such rules of conduct that would directly govern
> intelligent-like external behaviour so to completely fool the judges at
> the Turing's test.
But that's EXACTLY what you suggest we do as well. You suggest we build a
machine which is able to take your comprehension tests and completely
"fool" the testers into believing it's intelligent. You are hung up on the
word "fool" which has no significant meaning in this context. Either the
machine acts intelligent or it doesn't. If it acts intelligent, we will
believe it is intelligent. It makes no difference how it manged to act
that way. You are hung up on the idea of "fooling" the Judge when in fact
he never even once used the word "fool".
The point about using a behavior test is to take the focus on "can a
machine think" away from worrying about internal conscious thought and
feelings which are all subjective and untestable and turning the focus onto
simple objective behavior. And whether you understand or not, this is
exactly what you suggested we do (test the behavior of your machine).
> And that is precisely what most of AI community has
> been doing since.
Yes, that is true. But Truing isn't the AI community. Do you know that he
killed himself a few years after writing that paper? He wasn't around to
show the AI community what it should have been doing.
Reminds me of the "Goodbye idiots!" quote from another thread here in c.a.p
today.
> I think you've seen this problem yourself and that's
> why you don't concern yourself with attempting to program any rules that
> would look like intelligent behaviour.
Yes, I strongly believe that the only way to create intelligence is to
hard-code learning skills into a machine instead of hard-coding
"intelligent behavior". However, because I think intelligence is learning,
I've redefined my meaning of "intelligence" to be "strong learning
behavior". Which means that if I build a machine with strong learning
behavior, I am in fact, hard coding intelligent behavior into the machine.
It's just a confusing semantic problem. In the end, if intelligence has a
mechanical cause, then whatever mechanics we have to build into a machine
to make it intelligent is the mechanism of intelligence.
> > Lots of people who talk about the "Turing test" have no clue what
> > Turning was talking about. You sound like one of them. Read his paper
> > and stop listening to all the fools that have no clue what Turning was
> > talking about.
>
> Obviously whatever Turing has said can be interpetated in many ways. If
> I really want to, I can interpetate him sayings in such way that it
> makes it appear like he has the exact same idea that I have. But I am
> afraid if that is the case, he is not making himself very clear, and
> then the problem is that great many people have gotten a completely
> wrong idea from his sayings.
Exactly. Turning died in 1954. The first use of the term "Turing Test"
happened long after that. The best reference I can find about it was
wikipeda that says the term was first well publicised by Aurthur C Clark in
"2001 a space odyssey" (1968). Turing wasn't around to keep people from
misunderstanding his intent from his 1950 paper. If he had in fact lived
longer, AI might have been solved 25 years ago. He was already personally
playing with learning algorithms in 1950.
> Anyway, I doubt this to be the case, because if Turing was indeed having
> the same idea that I do, he would not have proposed such a thing as
> Turing's test at all.
Right. But now that you have proposed it, it's the right answer?
The point, whether you get it or not, is that he proposed it to get people
to stop debating the emotionally charged issue of whether a machine could
be conscious and to instead, simply focus on behavior. This is something
which is still emotionally and wrongly debated almost every day here in
c.a.p. It's the stuff which Glen keeps telling people not to do, and which
people keep calling him a fool for saying it 56 years after Turning
understand how wrong it was to waste time debating consciousness when in
fact behavior, is all there is to the problem of AI.
> When we build the first intelligent machine, we
> will not need a Turing's test to find out. In fact, it is madness to
> think that the first intelligent machine will even communicate in
> english or any human-like language. It will be more like a cat or a dog
> or an ape at most. ("Intelligence" here is the component pre-requisite
> of consciousness) We will know because it it learning and applying
> whatever it has learned into new situations in such a creative manner
> that was not programmed in. None of its external behaviour is not
> programmed in.
Sure. But I've already built intelligent machines. The problem is that no
one else will believe it until I use the same type of technology to produce
a human-level intelligent machine (or at least a cat or dog level machine).
> >>Sure, but if I had programmed an algorithm that was mimicing behaviour,
> >>I wouldn't consider passing Turing test to have any value at all.
> >
> > You should. Have you ever tried to do it? Have you played the Turning
> > test with a program written to attempt to mimic intelligence? People
> > have tried real hard to make programs that simply won the Turning test
> > by fooling people. No one has come close because it can't be done.
> > Either you create real intelligence, or it will fail the Turning test.
>
> I mean I wouldn't consider it to have any value at all in implication
> whether or not the system has consciousness.
Yeah, other people say that as well. They are wrong. Behavior is all we
have to judge with.
> > Do you know there is a yearly contest with a $100,000 prize to the
> > first program that can create a machine to pass the Turing test? If
> > mimicking intelligence is so easy, why hasn't someone done it and
> > picked up that $100,000 prize?
>
> I'm not saying it's easy, I'm saying it's impossible. I'm saying it is
> the wrong method in building intelligence. The attempts to win that
> prize have nothing to do with trying to create conscious intelligence.
> They are just technical masturbation.
Yes, I agree that most the entries submitted to win that prize or most
attempts to build an entertaining chat-bot are headed in the wrong
direction. However, most people who wrote those, either knew they weren't
creating intelligence before they even started, or figured out just how
unintelligent their creation was as they created it. And some of them, are
fixing their mistakes, and starting to create things that are actually
becoming more intelligent. They are working on chat-bots that learn how to
chat from experience, instead of having the programmer hard-code in
thousands of rules about how the machine should respond. I believe the
second place software from last year was a program that learned how to chat
from experience instead of having all it's language skills simply
hard-coded by programmer. If they keep working on that, they will end up
creating intelligent machines.
The problem is that the nature of a chat-bot contest makes the programmers
focus on the wrong problem. But in the end, if it's going to pass the
Turing test after an extended and unlimited set of interactions, it's going
to have to have real intelligence.
The real problem with the Turning test is what you said above - it's too
hard. That's because to act like a human it doesn't just have to be
intelligent, it has to have had a fairly human-like existence so it can
answer all the common-sense things that humans knows from their experience
- stuff that for the most part, you can't even find talked about on the
internet because it's such a common sense fact. The Turing test for
example won't help us know if we have manged to create dog intelligence
like you said. But the main point of the Turning test - that we should
ignore mentalist issues like consciousness and simply test behavior, is
exactly the answer to how we will know when we have dog intelligence. We
will compare the behavior of the machine to the behavior of dogs and see if
the machine is as good, or better, at all tests we can dream up to test
what we want to call "dog intelligence".
> > Have you read the transcripts to see how "intelligent" the best
> > programs are these days? Try it:
> >
> > http://loebner.net/Prizef/2005_Contest/Transcripts.html
> >
> > They are complete jokes. You can't mimic intelligence. The only way
> > for a program to pass an extended Turing test, is to be intelligent.
>
> Yeah sure, and another joke is that sometimes the judges mistake real
> persons to be machines. I don't know why people even bother with that
> test.
>
> I think it would be interesting if people tried to crack some simpler
> "semantical" task first. Like just a machine learning to move around in
> an environment fluently, or something else that simple biological
> organisms are capable of.
? Have you no clue that they were doing that back in the 50s and that they
have continued to do that up until today?
> For example, it would be interesting apply your networks so to control a
> physical body in a virtual environemnt.
Yeah, I was doing that in the 70s. I've been doing it ever since. I
typically don't get more complex than a rat running around a 2D maze
looking for food because it's easy to display that type of environment even
on a character based CRT which is what I had easy access to in the 70s.
And over the years, as I better appreciate the core problems of interacting
with an complex environment, my tests and my virtual environments have only
gotten simpler and simpler. As I get networks that work better in the
simplest and most fundamental problems, then I expect to advance to ever
increasing complex simulations. There's really no need to build robots if
you can't make your algorithms do intelligent things in simple simulated
environments.
> Just have some sort of thing
> with senses and limbs, and allow it to move its limbs freely. And then
> put some evolutionary pressure for it to learn to move around. To learn
> to jump and to climb and whatever is necessary in any environment.
> Perhaps even stack boxes.
Finding food is a good motivation. You can test to see how good it as
finding food to see how intelligent it is. If it can solve simple 2D
problems, then sure, simulating a 3D environment and hiding food under
boxes and seeing if learns to lift boxes to search for food would be a
great test for intelligence. I'm not there yet.
> Basically try to generate pressure for such a behaviour to arise that
> would "understand" in some way the meaning of the motion of the limbs,
> and how the world where it lives in works. And how to apply the motion
> of limbs to reach places. Or whatever is declared to be "good" by the
> critic (getting to "food", etc...)
Yeah, exactly. Like I said. I've been working on that problem for about
30 years now. Other people were working on the same problem 20 years
before I started. Marvin Minsky played with trying to get his SNARC
hardware to learn to find food in a virtual maze using trial and error
learning as part of his Ph.D. work back around 1953. It's considered one
of the first neural computers. He finds time to post here every once in a
while. Where have you been all this time?
You are suggesting stuff that was suggested and experimented with 50 years
ago and you are saying the guys that did it 50 years ago were clueless and
you now see the light. I don't think you grasp how much you still have to
learn about just what you think you are talking about.
I do think you are on the right track. I just think you are putting down
alternatives, and previous work that you don't understand as well as you
think you do.
As the saying goes, those that don't study history are destined to relive
it. I definitely see that happening with you.
The good thing about this constant cycle of life however is that each new
generation that relives the past, does so from a slightly different
perspective. And sometimes, that slightly different perspective is what
allows the new guy to make the breakthrough which the past generation
failed to make.
> -Anssi
Lucky
>thought in. She said that when she was at work she thought in English and
>when she was around Chinese-speaking people outside of work she thought in
>Chinese. Which was her primary language, under your scheme?
>It seems to me that language is context/use sensitive, and not as sharply
>defined as you claim it is.
Perhaps she lied. Perhaps you misunderstood her. Perhaps you heard
what you wanted to hear. Does the same thing apply to me? Yes. The
difference is that I am not relying on one person, like you.
-Jason
Lucky
>expect we'll just have to take your word for it. Seems we just have to
>take your word for pretty much everything you say.
Not true. Please don't transform your weak ability to utilize logic
into "you provide no proof", etc.
>> You realize that "Chinese" is
>>a human invention?
>Not really.
Um...Yes. Do your research. It is a human invention.
>> A human being will behave in the manner of a human being.
>Yeah that's real bright.
I have "dumb it down" for you.
>>He will make many mistakes at first, as knowledge is gained he will make fewer
>>and fewer. He will never reach perfection, but that's o.k. since no machine will
>>either.
>I think we can safely conclude you will indeed never reach perfection.
It is not even a goal of mine.
>>Are you still caught up in your belief about there being a "universal language"
>>involved in this arugement? If so, then yoiu need to read the thread "Observation
>>and Assumptions in the Chinese Room Arguement", where I use logic todisprove
>>such notion. The system is using the pre-existing abilities of the
>>individual componets in order to communicate. No spoken language is used.
>My, my, seems you're better at disproving things than proving them.
My, my, seems like you couldn't think of a comeback.
>>Universal Conspiracy theory? What the hell are you talking about? Look,
>>I don't feel you are correctly comprehending the information communicated to
>>you. It's most likely an ego thing. Surrender your ego and you should have no
>>further difficulties.
>Well Curt has his linguistic universal conspiracy theory and Glen his
>cognitive universal conspiracy theory. Now it seems we have your own
>variation with fundamental flaws to explain why your theories are true
>but no one especially cares to believe what you say.
1.) What Flaws?
2.) "No one expecially cares to believe what you say"? That's
weird....I don't recall anyone signing over their individuality to you.
Especialy since logic and intelligence is not your strength.
>I much prefer to judge your words with my ego since you've just
>explained to us that you won't prove what you say yet expect us to
>believe what you say anyway.
>~v~~
Oh, I see. You can't follow me, so you assume everyone else shares
your cognitive weakness. Unfortunately I don't feel this is the case.
I feel that many a person here is highly intelligent. My question to
you is, "Why are you interested in Arificial Intelligence, when you've
yet to master the human variety?"
-Jason
Anssi Hyytiainen
> Yes. I will ask you again. Can you comprehend consciousness.
> Answer with an initial "Yes" or "No", followed by a simple
> explanation.
Yes; I can "comprehend" consciousness in the same sense that I can
comprehend anything; I can have "an idea" about what consciousness is.
But
No; I cannot comprehend it in the same sense of "understanding it
completely". I cannot comprehend anything in that sense; I cannot truly
and fully understand it as if there was any "true" way to express what
consciousness is. I can only understand ideas semantically and their
meaning depends on whatever meaning I have placed on everything else in
my worldview.
Also, in the "no"-answer "comprehend" has been used with slightly
different meaning than in "yes"-answer. It is a bit difficult term in
that people often use it in the sense of someone really grasping some
idea "fully". But whatever they are grasping, it is still just an
"idea". (And whether or not they have really grasped it "fully", well
that's always up to debate)
> My theory suggests that comprehension is a process. I believe we
> have no choice but to comprehend information that has gone through
> such process, even if it is entirely wrong. Have you noticed how we
> always comprehend something, even if it is completely wrong!
Yeah, that sounds familiar. Without process there is nothing, that goes
for everything in the universe actually, and it is important to remember
when talking about the nature of consciousness. In any case, we can say
that information coming in through our senses affect the brain in some
way, and it is really up to the semantical learning to decide how we
experience this information.
We don't experience electricity or patterns, we experience things in
ways that have turned out to be "effective" in evolutionary pressure
both over generations and in our own head. The way bats "experience"
their sonar signals is nothing like how we experience sounds, I think we
can be pretty sure of that. They experience them in an "effective" way.
(If ever I can use the word "experience" for an animal that probably
doesn't have the capacity to ever figure out that it exists)
> So, can we comprehend consciousness? Yes, it's just that our
> comprehension will be completely wrong since it is not based on
> external sensory input. The comprehension that will arise is simply
> self-deception. "The Deceptive Consciousness".
Yeah, one could put it like that.
> So my next question would be, are we capable of "seeing" beyond the
> comprehension process?
If you mean, are we capable of comprehending how "the process" springs
consciousness exactly, my guess is still "no". We can have an idea about
it, but it is always going to be slightly unsatisfactory.
> You did, sorry! :) Oh wait! Could our two minds be inhabiting the
> same body?
>
> AAAAHHHHHHHHHHHHHHHH!!!!
Who knows. I haven't been conscious all the time. "Had I been sleeping
longer and longer??"
-Anssi
Anssi Hyytiainen
> I once asked a co-worker of mine who was born in Macau what language she
> thought in. She said that when she was at work she thought in English and
> when she was around Chinese-speaking people outside of work she thought in
> Chinese. Which was her primary language, under your scheme?
>
> It seems to me that language is context/use sensitive, and not as sharply
> defined as you claim it is.
I'm a finnish speaking person, but I can tell you that I am often
"thinking" in english. I am thinking in english as I am typing this.
Sometimes I can think of something I want to say only in english, and it
can be a struggle to figure out what the same thing is in finnish, and
sometimes vice versa. Sometimes when I have thought of something in
finnish, I write an english sentence with finnish structure (and it may
end up meaning completely the opposite of what I meant ;)
Anyway, as often as I'm thinking "in some language", I may be just
thinking about mechanics, or some things that are not like language at all.
Mostly we think in some language when we are thinking about how to
communicate some idea to someone. For example, when I'm just thinking
about an idea of consciousness, and I am thinking about how I could
express the idea in here, I am thinking about it in english even if I
don't intend to write anything down in days. And if I am to explain some
of that to someone in finnish, it can be a real struggle to translate
some words/concepts that I have thought about.
So in that sense language is context-sensitive; it depends on whom do
you want to communicate an idea to. If it doesn't matter, I think it is
pretty much 50/50 for bi-lingual people (I don't consider myself
bi-lingual btw; I still find it easier to communicate in finnish in
everyday situations), and often times you can think in multiple
languages in each "sentence"; some words in finnish and some in english.
-Anssi
Anssi Hyytiainen
> Anssi Hyytiainen <ans...@nic.fi.ANTISPAM> wrote:
>> That he thinks it would be easier to program a child's brain than
>> an adult's brain implies that he thinks one must program the
>> knowledge, the rules of (conscious) conduct, into the system.
>
He said it would be easier to program a child's brain. There's no reason
to assum the algorithm to be any different between them.
Doesn't that suggest he was thinking about programming the knowledge of
the world - that we are conscious of - into the system?
> Do you think you could program a learning machine without writing any
> code or building any hardware? Don't you think that whatever code
> you end up writing, or hardware you end up building is in fact the
> "mechanism of the child-brain"?
...which is the same mechanism that an adult brain would have. So what
are you saying?
> It's true that 56 years ago, he didn't understand as much as we do
> now.
>
> But what is irritating me is that you are saying all these negative
> things about Turning, and what he suggested we do 56 years ago, then
> you turn around and suggest that what we should do instead, but you
> are too
naive to
> even understand that what you are suggesting, is exactly what he
> suggested in that paper 56 years ago.
Sure, Turing was the first person who was starting to grasp the
implications of digital information processing. That is not to a reason
to not criticize his ideas of consciousness. The nature of consciousness
doesn't have to do with digital information processing alone, it it
something that Darwin and the field of biology are closing in as well.
I am not saying that he was a moron to not have figured out these things
immediately, I am just saying that reading his papers as if they were
describing the nature of intelligence (in the sense of something that
can become conscious), is rather like reading Newton's papers on the
nature of light. Sure, Newton was probably the most important physicist
who ever lived, that's not the reason to think his papers are of more
value now that newer updates.
And like I said, I do understand that at the heart of it all, the "true
nature of intelligence" is what Turing is basically saying, but there is
a clear difference in what he is implicating and what I think
intelligence requires. And the difference is the same which is there for
the traditional view of what people think "machine intelligence" is. I
would like to stress that when I just use the word "intelligence", it is
never meant to refer to an algorithm that is trying to mimic intelligence.
I basically agree with whatever Jeff Hawkins is expressing in the
beginning of his book, about how the whole AI industry was trying to
reach intelligence through a wrong route completely, the mimicry route,
and a lot of that is because of lack of consideration, and I believe to
an extent, due to Turing's gospels.
Do you understand for example, how you cannot just dismiss Searle's
view? How he has in many ways just as much "truth" in his sayings as
Turing does?
> But that's EXACTLY what you suggest we do as well. You suggest we
> build a machine which is able to take your comprehension tests and
> completely "fool" the testers into believing it's intelligent. You
> are hung up
on the
> word "fool" which has no significant meaning in this context. Either
> the machine acts intelligent or it doesn't. If it acts intelligent,
> we will believe it is intelligent. It makes no difference how it
> manged to act that way. You are hung up on the idea of "fooling" the
> Judge when in fact he never even once used the word "fool".
"It would deliberately introduce mistakes in a manner calculated to
confuse the interrogator."
He is talking about implementing rules of conduct directly for the
external behaviour a lot. Not about how the rules of conduct of learning
behaviour govern the way the machine can come to have its own ideas
about the world and meaning of things.
> The point about using a behavior test is to take the focus on "can a
> machine think" away from worrying about internal conscious thought
> and feelings which are all subjective and untestable and turning the
> focus
onto
> simple objective behavior. And whether you understand or not, this
> is exactly what you suggested we do (test the behavior of your
> machine).
In the case of intelligence, we are really only testing whether the
machine can learn anything at all, or if it's just going to "lie down on
the ground twitching."
In order to build truly capable learning machines, one should probably
expose the learning algorithm itself to evolutionary process.
You can't take all your focus away from "can a machine think"-types of
questions though. It is important to ask that question so to get an idea
about what explicit mechanisms have to do with intelligence, or rather,
how can we be intelligent while laws of physics aren't? I think I have
answered that question, and it is very beneficial to have an idea of how
a machine can indeed think, rather than just believe that it can and
push to some direction and see what happens.
> Yes, I strongly believe that the only way to create intelligence is
> to hard-code learning skills into a machine instead of hard-coding
> "intelligent behavior". However, because I think intelligence is
learning,
> I've redefined my meaning of "intelligence" to be "strong learning
> behavior". Which means that if I build a machine with strong
> learning behavior, I am in fact, hard coding intelligent behavior
> into the machine. It's just a confusing semantic problem. In the
> end, if intelligence has a mechanical cause, then whatever mechanics
> we have to build into a machine to make it intelligent is the
> mechanism of intelligence.
Yes, we two have the exact same idea of what intelligence is, and I've
been trying to express that it is a kind of semantical problem when I'm
saying where Turing went wrong, since after all is said and done about
the nature of semantics, it does work on mechanical basis only, and that
was the central point of my counter-argument to Searle; that true nature
of semantics is not anything fundamentally "semantical", rather
semantics are also mechanical.
The above assertion is utter non-sense unless one has gone through this
path of redefining the meaning of intelligence to be learning behaviour,
and a specific type of learning behaviour at that.
>> Anyway, I doubt this to be the case, because if Turing was indeed
>> having the same idea that I do, he would not have proposed such a
>> thing as Turing's test at all.
>
> Right. But now that you have proposed it, it's the right answer?
I don't propose any test that is anything alike with Imitation Game.
> Sure. But I've already built intelligent machines. The problem is
that no
> one else will believe it until I use the same type of technology to
produce
> a human-level intelligent machine (or at least a cat or dog level
machine).
Well, after you've reached "intelligent" behaviour, there's still some
road to such a behaviour reaching consciousness. Just like in animal
kingdom.
It sholdn't be heard for people to swallow that traces of truly
intelligent behaviour can be seen already. They are present in all
behaviour that is shaping through an evolutionary process. I remember
hearing that they have succeeded in making learning systems fly
airplanes in military simulators. They couldn't do anything fancy with
their capacity, but they could keep the plane in the air all on their
own. That is, learning mechanism that wasn't build to be used in flight
simulators, it was just some general purpose learning system. I think it
was the same that was used in the game "Creatures", where it governed
the behaviour of some fluffy animals jumping around(?)
[Value of passing Turing's test]
>> I mean I wouldn't consider it to have any value at all in
>> implication whether or not the system has consciousness.
>
> Yeah, other people say that as well. They are wrong. Behavior is
> all we have to judge with.
I wouldn't if I had built a system mimicing external behaviour directly.
Obviously I would if I had built a learning system. I believe my own
consciousness is completely mechanical at heart. I thought I made that
pretty clear.
>> I think it would be interesting if people tried to crack some
>> simpler "semantical" task first. Like just a machine learning to
>> move around in an environment fluently, or something else that
>> simple biological organisms are capable of.
>
> ? Have you no clue that they were doing that back in the 50s and
> that
they
> have continued to do that up until today?
Well they haven't done that, or if they have, they haven't used it for
any robot have they? I have only seen these cumbersome things that keep
falling down if they come across anything that hasn't been programmed
in. Like staircases.
>> For example, it would be interesting apply your networks so to
>> control a physical body in a virtual environemnt.
>
> Yeah, I was doing that in the 70s. I've been doing it ever since. I
> typically don't get more complex than a rat running around a 2D maze
> looking for food because it's easy to display that type of
> environment
Oh, I wasn't talking about path finding behaviour, I'm talking about a
system that is actually deciding which "muscles" to move and how, so to
be able to move in an environment, and so to be able to move in any kind
of arbitray environment. One that can figure out how to use stairs or
ladders when it sees them for the first time, or one that can figure out
how to jump over chasms, and one that is as error-prone in its motions
as any animal is, but which can figure out how to then get back on its
feet on its own just like any animal.
Just out of curiosity, what kinds of senses are you blessing your
networks with? In what form do they see the maze around them?
even
> on a character based CRT which is what I had easy access to in the
> 70s. And over the years, as I better appreciate the core problems of
interacting
> with an complex environment, my tests and my virtual environments
> have
only
> gotten simpler and simpler. As I get networks that work better in
> the simplest and most fundamental problems, then I expect to advance
> to ever increasing complex simulations. There's really no need to
> build robots if you can't make your algorithms do intelligent things
> in simple simulated environments.
Sure. I would imagine just coming up with "standing" behaviour would
take a while for a learning machine. Especially if we are talking about
bi-peds. I'm thinking that if you built some sort of spider-like thing,
you would probably first see some sort of dragging behaviour if you
tried to make it walk across a room onto piece of food.
-Anssi
Lester Zick
comp.ai.philosophy wrote:
>>Of course you don't. You provide plenty of claims but no proof. I
>>expect we'll just have to take your word for it. Seems we just have to
>>take your word for pretty much everything you say.
>
>Not true. Please don't transform your weak ability to utilize logic
>into "you provide no proof", etc.
So you do provide proof for your claims? Like the rest of posts to
which you reply proof seems to have gone missing from your replies.
>>> You realize that "Chinese" is
>>>a human invention?
>
>>Not really.
>
>Um...Yes. Do your research. It is a human invention.
No shit? Are you perchance familiar with the concept of hyperbolic
irony?
>>> A human being will behave in the manner of a human being.
>
>>Yeah that's real bright.
>
>I have "dumb it down" for you.
"AI for Dummies"? I love it.
>>>He will make many mistakes at first, as knowledge is gained he will make fewer
>>>and fewer. He will never reach perfection, but that's o.k. since no machine will
>>>either.
>
>>I think we can safely conclude you will indeed never reach perfection.
>
>It is not even a goal of mine.
Who cares?
>>>Are you still caught up in your belief about there being a "universal language"
>>>involved in this arugement? If so, then yoiu need to read the thread "Observation
>>>and Assumptions in the Chinese Room Arguement", where I use logic todisprove
>>>such notion. The system is using the pre-existing abilities of the
>>>individual componets in order to communicate. No spoken language is used.
>
>>My, my, seems you're better at disproving things than proving them.
>
>My, my, seems like you couldn't think of a comeback.
Undoubtedly why I didn't reply.
>>>Universal Conspiracy theory? What the hell are you talking about? Look,
>>>I don't feel you are correctly comprehending the information communicated to
>>>you. It's most likely an ego thing. Surrender your ego and you should have no
>>>further difficulties.
>
>
>>Well Curt has his linguistic universal conspiracy theory and Glen his
>>cognitive universal conspiracy theory. Now it seems we have your own
>>variation with fundamental flaws to explain why your theories are true
>>but no one especially cares to believe what you say.
>
>1.) What Flaws?
You?
>2.) "No one expecially cares to believe what you say"? That's
>weird....I don't recall anyone signing over their individuality to you.
> Especialy since logic and intelligence is not your strength.
And especially since spelling isn't yours.
>>I much prefer to judge your words with my ego since you've just
>>explained to us that you won't prove what you say yet expect us to
>>believe what you say anyway.
>
>>~v~~
>
>Oh, I see. You can't follow me, so you assume everyone else shares
>your cognitive weakness.
Hey you're the one who's bitching.
> Unfortunately I don't feel this is the case.
>I feel that many a person here is highly intelligent. My question to
>you is, "Why are you interested in Arificial Intelligence, when you've
>yet to master the human variety?"
But you have?
~v~~
Allan C Cybulskie
> >I once asked a co-worker of mine who was born in Macau what language she
> >thought in. She said that when she was at work she thought in English
and
> >when she was around Chinese-speaking people outside of work she thought
in
> >Chinese. Which was her primary language, under your scheme?
>
> >It seems to me that language is context/use sensitive, and not as sharply
> >defined as you claim it is.
>
> Perhaps she lied. Perhaps you misunderstood her. Perhaps you heard
> what you wanted to hear.
The instant you start relying on these sorts of replies to an argument, it
is clear that your argument has been defeated. Basically, you have no
answer. And all I need is ONE counter-example to show that your general
view is not the case ...
Allan C Cybulskie
"Anssi Hyytiainen" <ans...@nic.fi.ANTISPAM> wrote in message
news:om1Kf.1772$Gk3....@reader1.news.jippii.net...
I totally agree with this. I think that this is probably most often the
case, which lends credence to ideas that conscious thought is at least in
general a preparation for a possible expression of the ideas. Admittedly,
when we are visualizing things, it is unlikely that that is the case ... but
then we are unlikely to think in language in those situations anyway.
norrin
Lucky wrote:
>
> Yes. I will ask you again. Can you comprehend consciousness. Answer
> with an initial "Yes" or "No", followed by a simple explanation.
>
Yes. Consciousness is the state of being awake.
A live person who is not conscious is asleep or in an unusual physical
state.
A person who's asleep may be dreaming and if so they may falsely
believe
they are conscious.
A person with closed eyes may be conscious. If they open their eyes
they
do not go to a new state, but usually they know more that way.
Maybe you're implying there's no difference between consciousness and
dreaming.
>
> My theory suggests that comprehension is a process. I believe we have
> no choice but to comprehend information that has gone through such
> process, even if it is entirely wrong. Have you noticed how we always
> comprehend something, even if it is completely wrong!
>
> So, can we comprehend consciousness? Yes, it's just that our
> comprehension will be completely wrong since it is not based on
> external sensory input. The comprehension that will arise is simply
> self-deception. "The Deceptive Consciousness". So my next question
> would be, are we capable of "seeing" beyond the comprehension process?
>
The only two places you used ? in your post is to ask a yes-no question
with an answer you already know, if you accept the previous statements.
Instead, I show that people can truthfully comprehend connsciousness.
Curt Welch
> Curt Welch wrote:
> > Anssi Hyytiainen <ans...@nic.fi.ANTISPAM> wrote:
> >> That he thinks it would be easier to program a child's brain than
> >> an adult's brain implies that he thinks one must program the
> >> knowledge, the rules of (conscious) conduct, into the system.
> >
> > You have no clue what he's talking about do you.
>
> He said it would be easier to program a child's brain. There's no reason
> to assum the algorithm to be any different between them.
>
> Doesn't that suggest he was thinking about programming the knowledge of
> the world - that we are conscious of - into the system?
He talks about it under the section: "7 Learning Machines"
He says: "Instead of trying to produce a programme to simulate the adult
mind, why not rather try to produce one which simulates the child's? If
this were then subjected to an appropriate course of education one would
obtain the adult brain. Presumably the child-brain is something like a
note-book as one buys it from the stationers. Rather little mechanism, and
lots of blank sheets"
How is that anything other than a learning machine which starts off with
nothing but "blank sheets" and then fills it all in from learning?
> > Do you think you could program a learning machine without writing any
> > code or building any hardware? Don't you think that whatever code
> > you end up writing, or hardware you end up building is in fact the
> > "mechanism of the child-brain"?
>
> ...which is the same mechanism that an adult brain would have. So what
> are you saying?
Huh? What are YOU saying? :) That's half in jest and half serious.
Learning machines can't start off empty. They start off with some default
initial behavior, and then modify their behavior over time.
It's like a computer with memory. It's impossible for the memory to ever
be "empty". It's always there and it always has some value and the value
of the memory in the computer always causes to the computer to start with
some initial behavior (which might not be any more interesting than reading
the first instruction out of memory and halting). But the point is,
learning machines can't be a blank slate even though it's ok at times to
think in those terms. So the point Turning was making that you would have
to program some initial starting behavior into the machine and program some
ability for it to learn from experience (change it's behavior based on
experience).
My point is how is any of Turning's ideas about building a Learning machine
any different from your ideas?
> > It's true that 56 years ago, he didn't understand as much as we do
> > now.
> >
> > But what is irritating me is that you are saying all these negative
> > things about Turning, and what he suggested we do 56 years ago, then
> > you turn around and suggest that what we should do instead, but you
> > are too
> naive to
> > even understand that what you are suggesting, is exactly what he
> > suggested in that paper 56 years ago.
>
> Sure, Turing was the first person who was starting to grasp the
> implications of digital information processing. That is not to a reason
> to not criticize his ideas of consciousness. The nature of consciousness
> doesn't have to do with digital information processing alone, it it
> something that Darwin and the field of biology are closing in as well.
Ah, ok. I just don't happen to believe that. I happen to believe that
"consciousness" is nothing but a very simple signal processing effect. I
don't pretend to have any proof of this by it's something I happen to
believe very strongly. I actually believe that a simple signal processing
machine will be conscious. I don't know if traditional computers have
enough information processing power to create human level consciousness. It
might take custom designed hardware in order to push enough data in the
right way to create human level consciousness, but in the end, I believe we
will find that consciousness is nothing more than a simple signal
processing algorithm.
In other words, this neat feeling we all have of being conscious is nothing
more than what it feels like to be the right type of signal processing
machine.
> I am not saying that he was a moron to not have figured out these things
> immediately, I am just saying that reading his papers as if they were
> describing the nature of intelligence (in the sense of something that
> can become conscious), is rather like reading Newton's papers on the
> nature of light. Sure, Newton was probably the most important physicist
> who ever lived, that's not the reason to think his papers are of more
> value now that newer updates.
Yeah, and you might believe there is more to consciousness than either I or
maybe Turing believed. I suspect that he believed the same thing I do.
> And like I said, I do understand that at the heart of it all, the "true
> nature of intelligence" is what Turing is basically saying, but there is
> a clear difference in what he is implicating and what I think
> intelligence requires. And the difference is the same which is there for
> the traditional view of what people think "machine intelligence" is. I
> would like to stress that when I just use the word "intelligence", it is
> never meant to refer to an algorithm that is trying to mimic
> intelligence.
Ok, so the bottom line is that I do happen to believe that human
intelligence and consciousness is nothing more than a simple machine
specified by a simple algorithm. Maybe you believe there is something more
complex to it. Most people do.
> I basically agree with whatever Jeff Hawkins is expressing in the
> beginning of his book, about how the whole AI industry was trying to
> reach intelligence through a wrong route completely, the mimicry route,
> and a lot of that is because of lack of consideration, and I believe to
> an extent, due to Turing's gospels.
I just finished his book tonight. I think I understand what Jeff is
thinking now. I think he believes like I do that intelligence and
consciousness is nothing more than the behavior of a simple machine
following a simple algorithm. He sees the algorithm as being his
memory-predictive stuff, I see it as being a bit more. But we both see all
the effects of consciousness as nothing more than what happens when you
build the right type of machine. He talks about that issue very explicitly
towards the end of the book when he brings up a conversation he had with
other scientists about consciousness one evening.
However, he has this perception-centric view of intelligence which totally
ignores the prime problem of AI - which is behavior. His theory seems to
be that the intelligence is created by this memory-predictive hardware
which creates our ability to understand the meaning in our sensory signals.
What outputs we then choose to produce using this understanding of the
structure of the sensory data seems to be almost unimportant to him. He
sees the output side of the problem is just some simple inverse of the
perception problem - which is totally illogical and I'll have to argue that
with him some day.
So, his problem with the Turing test is that it caused people to focus on
behavior (which he seems to mean - the output of the black box), and miss
what he sees as the real solution to AI, which is this perception hardware
which happens inside the black box, independent of the box producing any
output at all. So he thinks he has solved AI (or at least found the
correct framework to explain it) by ignoring the output and focusing on
what is happening inside. So while everyone else was wasting time trying
to get their programs to produce the right output, he feels he has found
the key to AI by focusing on what is happening inside.
I think he's missed the mark big time. I think for the most part, I agree
with everything he says, but he's only got a handle on half the problem,
and left the most important part - the output, totally unsolved.
> Do you understand for example, how you cannot just dismiss Searle's
> view? How he has in many ways just as much "truth" in his sayings as
> Turing does?
No, Searle's and idiot. :) (ok, a smart idiot). What took me forever to
figure out was to understand how anyone couldn't just instantly dismiss it.
But the bottom line is that some people feel strongly that consciousness
can't be explained as an algorithm. I just happen to believe very strongly
that it can be. What I later understood is that if you believe what I
believe, the Chinese Room looks like the most stupid argument anyone could
make. It's so outright stupid, I would be embarrassed for a 1st grader
trying to use that logic to make an argument. But if you happen to believe
the other way, it looks like a brilliant argument. The truth is that it's
not an argument at all. It's just an ink blot test that allows you to see
what you already happen to believe. Anyone that doesn't see that, is just
too invested to their own beliefs to see the argument from the other side.
I very much pick one side for my own personal belief, but I now understand
why others pick the other side of the debate.
The truth is we don't have the evidence to show which side is right, and
the Chinese Room doesn't add any more evidence or support for either
position. It's not stupid as I first thought, it's just useless.
> You can't take all your focus away from "can a machine think"-types of
> questions though. It is important to ask that question so to get an idea
> about what explicit mechanisms have to do with intelligence, or rather,
> how can we be intelligent while laws of physics aren't?
> I think I have
> answered that question,
Well, I too have answered it. My answer is that the laws of physics are
intelligent.
> and it is very beneficial to have an idea of how
> a machine can indeed think, rather than just believe that it can and
> push to some direction and see what happens.
The bottom line is that if anyone knew the answer, we would know which
direction to push. But since no one knows the answer, we each make up are
own best guess to the answers, and then push in that direction. It turns
out, that there are many ways to explain these effects, and each of these
different explanations suggest we push in a different direction. So, we
each choose the answers we like best, and push in that direction (or
actually, most of us push in some direction we like best - and then make up
answers to justify the direction we are headed). No one will know if their
answers are correct until someone gets to the end and has a working
intelligent machine that is so damn intelligent and human-like that no one
will deny what it is after they spend enough time with it. It might have
to be 10 times more intelligent than any human before people will actually
accept that it is intelligent however (the "robot" tax).
> > Yes, I strongly believe that the only way to create intelligence is
> > to hard-code learning skills into a machine instead of hard-coding
> > "intelligent behavior". However, because I think intelligence is
> learning,
> > I've redefined my meaning of "intelligence" to be "strong learning
> > behavior". Which means that if I build a machine with strong
> > learning behavior, I am in fact, hard coding intelligent behavior
> > into the machine. It's just a confusing semantic problem. In the
> > end, if intelligence has a mechanical cause, then whatever mechanics
> > we have to build into a machine to make it intelligent is the
> > mechanism of intelligence.
>
> Yes, we two have the exact same idea of what intelligence is, and I've
> been trying to express that it is a kind of semantical problem when I'm
> saying where Turing went wrong, since after all is said and done about
> the nature of semantics, it does work on mechanical basis only, and that
> was the central point of my counter-argument to Searle; that true nature
> of semantics is not anything fundamentally "semantical", rather
> semantics are also mechanical.
Well, some people just don't accept that semantics is mechanical because
they believe true semantics is a result of consciousness and they see
consciousness as non-mechanical almost by definition. In short, some
people strongly believe that we are just not mechanical machines - we are
something beyond that. They don't know what we are, but they know in their
heart that we are not just machines - and that you would never create a
conscious machine just by writing software.
I think computers are in fact already conscious. They are consciously
aware of me typing on the keyboard for example. They make conscious
decisions to print a document in response to me asking it to do that. I
just think that computers don't have the same level of consciousness a
human has. But most people are taught to believe that the above ideas are
wrong - they are taught from a very young age that humans are special
(animate) and not like the machines (inanimate). Some people believe what
they were taught as a kid. I've rejected what I was taught because it's
just illogical and doesn't fit the actual facts available to us. I believe
people like Searle (and more than half the population) just don't have what
it takes to get best their training and see the truth.
I of course could be dead wrong. I understand that. But that's my story
and I'm sticking to it. :)
> The above assertion is utter non-sense unless one has gone through this
> path of redefining the meaning of intelligence to be learning behaviour,
> and a specific type of learning behaviour at that.
Yeah, but if that's is what intelligence is, then you find you really can't
draw a line between intelligent, and non-intelligent things. In the end,
you are forced to say that rocks are intelligent - just not as intelligent
as humans. Only human-like intelligence requires a specific type of
learning machine design.
The current problem is that humans are so far separated from the next most
intelligent thing on the planet, that we are not left with the impression
that we are part of a large continuum. This makes people believe there is
some type of hard line between humans and everything else. As we develop
intelligent machines, the line will vanish. The world will be full of
intelligent machines of all shapes and sizes and humans will just stop
looking so special - if anything, they will start to look very unspecial.
The only thing that will make them special in the distant future is the
unique role they played in the history of the earth. But in another
million years, even the role that humans played for a few thousand years
will probably be seen as nothing but a minor footnote in the history of the
universe.
> >> Anyway, I doubt this to be the case, because if Turing was indeed
> >> having the same idea that I do, he would not have proposed such a
> >> thing as Turing's test at all.
> >
> > Right. But now that you have proposed it, it's the right answer?
>
> I don't propose any test that is anything alike with Imitation Game.
Didn't you say you thought the way to test intelligence was to give the
machine a comprehension test and see what type of answers it produced? To
me, this is exactly what the Turning test, and the imitation Game is all
about.
The hidden point behind the Turning test is the belief that consciousness
is nothing special - so you don't need to test for it separately from the
physical behavior of the machine. The behavior of the machine is all there
is (to people that believe like I do). If it behaves correctly, then that
is all we need to know about it and that is all there is that we could ever
possibly know about it. The fact that people believe subjective knowledge
is private and not available for objective observation is yet another false
conclusion based on the lies taught to us. Everything is objective and
that's all there is. If you get the objective facts correct, you get the
subjective facts correct at the same time. This is the point the Turing
test is making without actually saying it.
> > Sure. But I've already built intelligent machines. The problem is
> that no
> > one else will believe it until I use the same type of technology to
> produce
> > a human-level intelligent machine (or at least a cat or dog level
> machine).
>
> Well, after you've reached "intelligent" behaviour, there's still some
> road to such a behaviour reaching consciousness. Just like in animal
> kingdom.
I don't think so. I think my computer is already conscious. I think the
only issue here is shaping it to make it more of a human like
consciousness.
> It sholdn't be heard for people to swallow that traces of truly
> intelligent behaviour can be seen already. They are present in all
> behaviour that is shaping through an evolutionary process. I remember
> hearing that they have succeeded in making learning systems fly
> airplanes in military simulators. They couldn't do anything fancy with
> their capacity, but they could keep the plane in the air all on their
> own. That is, learning mechanism that wasn't build to be used in flight
> simulators, it was just some general purpose learning system. I think it
> was the same that was used in the game "Creatures", where it governed
> the behaviour of some fluffy animals jumping around(?)
>
> [Value of passing Turing's test]
> >> I mean I wouldn't consider it to have any value at all in
> >> implication whether or not the system has consciousness.
> >
> > Yeah, other people say that as well. They are wrong. Behavior is
> > all we have to judge with.
>
> I wouldn't if I had built a system mimicing external behaviour directly.
> Obviously I would if I had built a learning system. I believe my own
> consciousness is completely mechanical at heart. I thought I made that
> pretty clear.
Ok. Some things you say seem to contradict other things you say so I'm not
always sure what your position is and how similar it is to mine or where we
might differ.
But in the end, if we are trying to build a car, and we build a chain-saw,
we will say, "hey that's not a car". We will know that because the
chain-saw doesn't behave like car. It's got different behavior.
If we build the right thing, it will behave correctly, and it will be the
right thing. If you build the wrong thing, it will behave incorrectly at
least at some times, and it will not be the thing we were trying to build.
If you get the behavior correct you will have succeeded in building the
right thing. It's impossible for it to behavior correctly, and be the
wrong thing. If I build a machine that acts like a car, then it IS a car.
That's all there is to it. It's impossible to build a machine that mimics
a car 100% and for it not to be a car. That's simply how we define what a
car is. It's a car by definition.
If you build a machine that acts like AI 100%, then it IS AI by definition.
The only way to justify that it's not AI, is if you are one of those people
that believe consciousness is some magic sauce mixed with the machine that
is somehow separate from the machine. It's just illogical to believe this
even though a large group of people do in fact believe it.
> >> I think it would be interesting if people tried to crack some
> >> simpler "semantical" task first. Like just a machine learning to
> >> move around in an environment fluently, or something else that
> >> simple biological organisms are capable of.
> >
> > ? Have you no clue that they were doing that back in the 50s and
> > that
> they
> > have continued to do that up until today?
>
> Well they haven't done that, or if they have, they haven't used it for
> any robot have they? I have only seen these cumbersome things that keep
> falling down if they come across anything that hasn't been programmed
> in. Like staircases.
Yeah, they don't look very life-like yet. Lots of people have built
learning machines but no one that I know of has got that fluid life-like
purposeful behavior we see in all animals. I would love to see someone
make that happen. I think my networks have a reasonably good chance of
doing that. And I think if anyone gets a machine to learn that type of
behavior on it's own without having an artist script the behavior into a
machine, it will be a big step forward in closing the gap between animate,
and inanimate objects. Until we substantially close that gap, all these
"fools" will keep hanging on to the false hope that conscious beings are
not just mechanical machines.
> >> For example, it would be interesting apply your networks so to
> >> control a physical body in a virtual environemnt.
> >
> > Yeah, I was doing that in the 70s. I've been doing it ever since. I
> > typically don't get more complex than a rat running around a 2D maze
> > looking for food because it's easy to display that type of
> > environment
>
> Oh, I wasn't talking about path finding behaviour, I'm talking about a
> system that is actually deciding which "muscles" to move and how, so to
> be able to move in an environment, and so to be able to move in any kind
> of arbitray environment.
Yeah, I know what you are talking about. And I very much want to see a
learning machine learn that. But right now, I'm working on much simpler
problems which I believe will lead us to what you and I both want to see
happen.
> One that can figure out how to use stairs or
> ladders when it sees them for the first time, or one that can figure out
> how to jump over chasms, and one that is as error-prone in its motions
> as any animal is, but which can figure out how to then get back on its
> feet on its own just like any animal.
Yeah, I'd love to see that. I'd love to create solar powered "bugs" that
were robust enough that they could survive on there own outside and had
motivations that made them a real bitch to hunt down and catch (they would
tend to avoid people). I'd just set them loose in the wild and see where
they would roam (maybe I'd put a radio tracker on them).
And if you could create bugs with enough power so it could fly - how cool
would that be. Maybe bugs that had the technology that they could tap into
the power of power lines and use that to power themselves. Or maybe bugs
that were just smart enough to find power outlets and plug in to get power?
Maybe they would have sense coils that would allow them to sense the
magnetic or static electric fields created by power lines and would have
teeth or claws strong enough to dig though walls to get to the wires so
they could tap in to the power? They would be power-rats chewing though
the walls of people's houses at night just to get to the power.
Lots of stuff like that I think will actually be possible with computers
not much more powerful than a standard laptop once we have the right
software.
And if you think computer virus are a problem now. Imagine what hackers
might create if they could create rat-like intelligence in virus software
and let it loose on the internet. Combine that level of intelligence with
some power to evolve and who knows what might breed on the internet. I
think there's going to be so really cool crap happening in the next few
decades.
> Just out of curiosity, what kinds of senses are you blessing your
> networks with? In what form do they see the maze around them?
Mostly very simple senses. In the past, I've played with grid worlds where
the AI-rat was located at some fixed position on the grid, and the maze
worked by having either walls, or open paths, in the four possible
directions (north, south east and west). The only behavior these rats
could make is to take a single step in one the directions, which would move
it to the next grid position - unless there was a wall there - which means
it would not move. The typical sense these rats had were to sense if there
was a wall in each of the 4 directions. There was food located at
different locations, and by moving to the grid location which contained
food, the AI-rat would get a reward and the food would vanish (sometimes
showing up randomly somewhere else in the maze).
Some of the rats could "smell" food in the 4 directions as well. Some
could "smell" food further away. You could create a sense that allowed it
to see down a hall and tell how long the hall was. There's really no limit
to what type of senses you can give it.
However, I've not played with worlds that advanced since the mid 80's.
That's because that's when I realized how none of my networks had the power
they needed to learn even trivial facts about how to find food in these
mazes. My networks could learn very simple behaviors, such as "don't walk
left if there is a wall to the left", or "walk left when there was food the
the left", and even learn things like "standing still isn't a good thing to
do to find food". But beyond that, they could not learn to produce a
sequence of behaviors (such as a simple N N E W sequence to get to the
cheese), and they could not learn to produce different sequences at
different times.
But playing with those made it clear that a big part of intelligence was
learning to produce the right sequence of behaviors at the right time.
Which also meant having a strong ability to correctly fit old answers to
new problems - to understand that a given context was similar to some past
experience and to correctly generate an appropriate response for the new
situation.
Playing with those simple worlds made it fairly clear what had to be done,
but trying to find a good way to implement it is what's been so hard. And
I don't need a complex grid maze to know if my designs have what it takes
to do interesting things. Very simple low level tests tell me tons about
whether I'm on the right path to creating rat-like behavior. So I've been
testing my ideas with very trivial problems to understand their power (or
lack of power).
> even
> > on a character based CRT which is what I had easy access to in the
> > 70s. And over the years, as I better appreciate the core problems of
> interacting
> > with an complex environment, my tests and my virtual environments
> > have
> only
> > gotten simpler and simpler. As I get networks that work better in
> > the simplest and most fundamental problems, then I expect to advance
> > to ever increasing complex simulations. There's really no need to
> > build robots if you can't make your algorithms do intelligent things
> > in simple simulated environments.
>
> Sure. I would imagine just coming up with "standing" behaviour would
> take a while for a learning machine. Especially if we are talking about
> bi-peds. I'm thinking that if you built some sort of spider-like thing,
> you would probably first see some sort of dragging behaviour if you
> tried to make it walk across a room onto piece of food.
Yeah, one idea I had to motivate something like a spider machine is to just
put light sensors on it and motivate it to try and maximize the amount of
light the sensors were picking up (or the inverse and try to minimize the
light - aka hide from the light). This would at first just move its legs
somewhat randomly, but at times, happen to do something to push the body a
bit more into the light (or just tilt it towards the light to make it pick
up a bit more light). The idea is that lots of these little accidents that
just happened to work to get it more light would be slowly shaped from
somewhat random leg movements to purposeful light-seeking walking behavior
over time. Just to see that happen would be very cool and a strong
indication that we were on the right path to creating intelligence. I
think with the right learning algorithm, a little light seeking spider-bot
would start to look very alive and very conscious to a lot of people. And
I think we are close to finding that learning algorithm.
Anssi Hyytiainen
> He talks about it under the section: "7 Learning Machines"
>
> He says: "Instead of trying to produce a programme to simulate the adult
> mind, why not rather try to produce one which simulates the child's? If
> this were then subjected to an appropriate course of education one would
> obtain the adult brain. Presumably the child-brain is something like a
> note-book as one buys it from the stationers. Rather little mechanism, and
> lots of blank sheets"
>
> How is that anything other than a learning machine which starts off with
> nothing but "blank sheets" and then fills it all in from learning?
You are right, he does say that.
Perhaps he should have made that section the central part of his
argument. I would have.
> My point is how is any of Turning's ideas about building a Learning machine
> any different from your ideas?
The section 7 doesn't seem any different now that I'm interpetating it
like you are.
>>Sure, Turing was the first person who was starting to grasp the
>>implications of digital information processing. That is not to a reason
>>to not criticize his ideas of consciousness. The nature of consciousness
>>doesn't have to do with digital information processing alone, it it
>>something that Darwin and the field of biology are closing in as well.
>
> Ah, ok. I just don't happen to believe that. I happen to believe that
> "consciousness" is nothing but a very simple signal processing effect.
Yeah yeah, I believe what you do. By asserting that the field of biology
is closing in on it as well, I'm refering to deeper understanding of
evolutionary processes.
Like there is the fallacy of our "mind" being somehow magically
intentional on its own, there was the fallacy of intelligent design;
everything around us seemed so purposeful as to be intentionally designed.
Then Darwin came by and described evolution as a mechanical process that
suddenly explains all the apparent intentionality around us.
Deeper understanding of evolutionary processes shows cleary that the
same process extends over all that exists in the universe. Stable
systems can do nothing but prevail, and they are all mechanical. From
the existence of quarks to the existence of conscious mind. I am not
sure if Darwin himself made this assertion about the nature of human
mind as something that mechanically evolves, but I suspect he may have
since the conclusion is so clear.
Field of biology is not bunch of people thinking that we have a soul.
They are very seriously looking at evolutionary processes. Dawkins makes
some assertions of mechanical mind in his book Selfish Gene.
Also I should mention that physics are closing in on the same thing.
Physics is largely about evolutionary mechanics, that is, thinking about
how certain structures or fundamental processes would manifestate the
things around us. And part of that is deeper understanding of the
fundamental processes going on in our brain.
In fact my conclusions about the nature of our consciousness draw pretty
heavily from the field of physics & evolutionary mechanics.
The basic structure of what I call "conscious process" needs to be of
proper "class". The so-called "intelligence" follows open-ended learning
mechanism that cannot understand anything in any other way but by
"guessing" basically.
It is tagging semantical meanings around with no basic guideline. Other
than "whatever meaning seems to "fit" due to all the other meanings, go
for it." There's no explicit base. If there was, there could be no
semantics. And so called "consciousness" follows when enough such
semantical learning has taken place.
> Yeah, and you might believe there is more to consciousness than either I or
> maybe Turing believed. I suspect that he believed the same thing I do.
I suspect you and I and Turing all believe the "same thing", but you and
I seem to look at the apparent structure of this mechanical process
little differently. More about that later on in this post ->
> Ok, so the bottom line is that I do happen to believe that human
> intelligence and consciousness is nothing more than a simple machine
> specified by a simple algorithm. Maybe you believe there is something more
> complex to it. Most people do.
I'm not one of those people.
> Well, I too have answered it. My answer is that the laws of physics are
> intelligent.
But not intentional? I know you are using the world "intelligent" in
much broader sense than I have been in this thread.
> I think computers are in fact already conscious. They are consciously
> aware of me typing on the keyboard for example. They make conscious
> decisions to print a document in response to me asking it to do that.
The difference in our views is that I find it very useful to draw a
clear line between the kind of "intelligence" you are talking about
above, and something that we could call "conscious".
Sure it is true that the line is completely semantical, but it is useful
nevertheless; We cannot be conscious of any actual mechanical processes
going on in our head, the same way as a conscious machine would not be
conscious of the actual algorithms springing its learning behaviour.
That is why I think calling everything conscious is not very helpful.
We could say that the relevant point to pick up from Searle's argument
is that we are not aware of the mechanics of our mind. What Searle fails
to see is that semantics are not some sort of magical intentional thing.
Semantics are the things we are aware of. Things we don't really KNOW to
be "the truth" the way that a computer "knows" 1+1 = 2. We only make
guesses and build "ideas". Our ideas are formed by an algorithm still,
but we should probably choose to use the word "consciousness" to refer
to such "loose" ideas only, so not to lead people to think that we
believe rocks have feelings and all that nonsense.
After all, it is just as true to assert that "pure intelligence" doesn't
exists in the universe at all. Just like pure intentionality doesn't
exist either (due to everything actually working through evolutionary
mechanics).
The point is that such a thing as "semantical mechanics" is not an
oxymoron (and HOW such a thing actually exists), even though many people
(incl. Searle) think it is.
So I really wouldn't use the word "conscious" when the computer "makes a
decision" to print a document in respons to such a command. If it made a
conscious decision it would need to understand semantically what the
command means (having learned the meaning of the command), and then it
would actually decide whether to print the document or not, depending on
everything else it knows about the situation.
> Yeah, but if that's is what intelligence is, then you find you really can't
> draw a line between intelligent, and non-intelligent things. In the end,
> you are forced to say that rocks are intelligent - just not as intelligent
> as humans. Only human-like intelligence requires a specific type of
> learning machine design.
All the intelligence we see in the animal kingdom is the same type of
intelligence as that of humans. Even the intelligence of worms. Class
distinction between animal behaviour and plant or rock behaviour is
useful, that's all. The learning capacity of most animals is just so
much less that most of them never come to learn enough, to make any
assumptions about the meaning of "existence" and "self". So I would even
draw a separate (semantical) line between intelligence and
consciousness, the argument being that there are species that are
"semantically intelligent" but not necessarily self-conscious.
Species-specific learning process (DNA) leads to -> instance-specific
learning process (nervous systems) -> semantical understanding ->
conscious experience
> The current problem is that humans are so far separated from the next most
> intelligent thing on the planet, that we are not left with the impression
> that we are part of a large continuum.
I was just wondering how many animals could actually learn little bit of
human language if they had a proper throat for it. That should reveal
people something about animals.
There is some simple understanding of "meaning" going on here between
food rewards and a "voice trick" these cats perform:
http://www.heavy.com/heavy.php?videoPath=/content/afv/flash_video/AFHV_TalkingCats
(takes a while to load)
The animals don't know the semantical meanings that WE have placed onto
the "words" these animals are saying, but there is "a" meaning
nevertheless. The "task" is still of the same class as language, or any
understanding of meaning. Evidently cats can understand the meaning of
the sound of a fridge being opened, or the relatively complex voice
pattern that is their own name. Some animals seem to understand
surprising amount of "meanings".
I'm sure that gorillas could learn to utter real words with
understanding their meaning if they had a proper throat. The same goes
for dolphins, who have in fact a developed a kind of language and each
have a unique name they use to address different individuals.
>>>>Anyway, I doubt this to be the case, because if Turing was indeed
>>>>having the same idea that I do, he would not have proposed such a
>>>>thing as Turing's test at all.
>>>Right. But now that you have proposed it, it's the right answer?
>>
>>I don't propose any test that is anything alike with Imitation Game.
>
> Didn't you say you thought the way to test intelligence was to give the
> machine a comprehension test and see what type of answers it produced? To
> me, this is exactly what the Turning test, and the imitation Game is all
> about.
Hmmm, could it be that you are mixing me with "Lucky"? I'm pretty sure
I'm not him, even if he is claiming so :)
And even if we live in the same brain, I am not responsible of his ideas :)
>>I wouldn't if I had built a system mimicing external behaviour directly.
>>Obviously I would if I had built a learning system. I believe my own
>>consciousness is completely mechanical at heart. I thought I made that
>>pretty clear.
>
> Ok. Some things you say seem to contradict other things you say so I'm not
> always sure what your position is and how similar it is to mine or where we
> might differ.
Yeah, it probably seems like I am contradicting myself on a regular
basis, but that's just because I have drawn a semantical distinction
between mechanical processes and semantical understanding that emerges
from a mechanical process. "Semantical mechanics."
What I mean with "semantical" is not exactly the same thing as when
Searle is talking about semantics, since he considers it to be somehow
fundamental. Fundamental intentionality or something like that. That's
nonsense.
>>Just out of curiosity, what kinds of senses are you blessing your
>>networks with? In what form do they see the maze around them?
>
> Mostly very simple senses.
My view of what "semantical mechanics" are, would suggest that
sophisticated senses should be hugely beneficial for the AI. The more
information you feed into the system, the more chances it has to draw
all kinds of meanings of these things.
For example, if I wanted to build a system that would learn to move
around in arbitrary environment, I would think the system would be
better off with;
1. proper visual information of its surroundings
2. orientation information of its own body (balance organ)
3. sense of touch (to allow it to learn to take into account the "feel"
of if there is something underneath its feet or not)
At least those three I think should be the starting point. If it just
has a sense of whether it can move to one of four directions or not,
there's not much meaning to be "picked up" from that.
Of course the good news are, that no matter how sophisticated the
sensory systems, you don't need to worry about anything but to feed all
the information, in whatever arbitrary form, into the system, and it
should pick up the meaning of it. Learn how to "see", for example.
-Anssi
Lucky
>comprehend anything; I can have "an idea" about what consciousness is.
>But
>No; I cannot comprehend it in the same sense of "understanding it
>completely". I cannot comprehend anything in that sense; I cannot truly
>and fully understand it as if there was any "true" way to express what
>consciousness is. I can only understand ideas semantically and their
>meaning depends on whatever meaning I have placed on everything else in
>my worldview.
>Also, in the "no"-answer "comprehend" has been used with slightly
>different meaning than in "yes"-answer. It is a bit difficult term in
>that people often use it in the sense of someone really grasping some
>idea "fully". But whatever they are grasping, it is still just an
>"idea". (And whether or not they have really grasped it "fully", well
>that's always up to debate)
O.K., follow me here:
We both agree that:
1.) We cannot think of that which we do not comprehend on some level.
If we do not possess certain information then the best we can do is
comprehend that we don't comprehend such.
2.) Comprehension does not have more than one meaning. It is
"understanding" in whatever level. Many people get hung up on the idea
that we must understand something in it's entirity, but this incorrect.
Therefore,
Comprehension = Thought/Thinking
Consciousness is the state of awarenss of our thoughts.
Therefore,
Consciousnous = Comprehension
Creat an A.I. that has the ability to comprehend and we immediately
get an A.I. that is conscious.
Still with me?
O.K. here's some more:
I propose "Comprehension" is a process, resulting from the structuring
of information stored by the brain. I have already prior made you
aware of my theory of such.
So according to my theory;
Comprehension is depedant on the information provided by sensory input
of the exterior environment.
Agree?
What happens when we try to comprehend the mind?
Information of "the mind" can never be pieced together from sensory
input from the exterior environment.
and
2 minds cannot inhabit the same brain while being consciously aware of
each other.
Therefore,
The mind cannot gather sensory input concerning itself, and it cannot
conciously break off a portion of itself so that it could act as an
"observer", therefore any theory of mind is flawed.
Comprehension is a process. Whatever information goes into this
process, is transformed into comprehension, even it if is completely
wrong.
Therefore,
When we attempt to think about the mind, any theory we form will be
flawed and incorrect. However, since comprehension is a process, we
will be able to understand such theory and accept it as fact.
Again here is a brief repeat:
1.) Conciousness = Comprehension
2.) The mind is dependant on sensory input from the external
environment.
3.) When trying to comprehend itself, the mind can not use sensory
input, therefore it is decieving itself.
4.) Since Comprehension is a process, any theory the deceptive mind
creates will be understood and accepted as fact, thought it will be
essentialy flawed.
-Jason
Lucky
>is clear that your argument has been defeated. Basically, you have no
>answer. And all I need is ONE counter-example to show that your general
>view is not the case ...
Read the rest of my reply. I also don't automaticly take someone's
word, like when you claim to sometimes think in English and sometimes
think in Finnish. I believe the mind is incredibly deceptive, to both
self and others. ;)
-Jason
Lucky
>with an answer you already know, if you accept the previous statements.
>Instead, I show that people can truthfully comprehend connsciousness.
How is that "truthfully comprehending"? So I would ask you define
"consciousness" and you would say being awake. I would then ask you
to define "awake" and you would respond "not being asleep", etc... and
around and around we go...
Sorry, I got better things to do.
-Jason
Lucky
>case, which lends credence to ideas that conscious thought is at least in
>general a preparation for a possible expression of the ideas. Admittedly,
>when we are visualizing things, it is unlikely that that is the case ... but
>then we are unlikely to think in language in those situations anyway.
You "totaly agree" because this only what you are capable of
comprehending. Since this is not based on sensory input, you are
performing unconscious self-deception.
-Jason
Lucky
>> machine a comprehension test and see what type of answers it produced? To
>> me, this is exactly what the Turning test, and the imitation Game is all
>> about.
>Hmmm, could it be that you are mixing me with "Lucky"? I'm pretty sure
>I'm not him, even if he is claiming so :)
>And even if we live in the same brain, I am not responsible of his ideas :)
LOL...yeah I stated that! However, I need to clarify that I didn't say
an "intelligence test", I stated give A.I. a "Comprehension Test",
which is completely different. Doesn't anybody remember taking a
Comprehension Test when they were young? For example, a Reading
Comprehension test?
For example, in a Reading Comprehension Test, questions would be asked
which require the reader to respond with answers not directly alluded
to in the text itself. The person must be able to "connect the dots"
in order to reply correctly.
For example:
"Somebody in the kitchen is celebrating a birthday today. Kirsten,
Mark, and their daughter Susan are sitting at the kitchen table. Mark
and Susan dismiss themselves from the table, complaining they are
tired. Mark and Susan then quietly sneak into the living room and
proceed to put party-hats on their heads."
Question: Who's birthday is it today?
Notice how difficult this would be for a machine? Notice how easily we
do it!
-Jason
Lucky
>word, like when you claim to sometimes think in English and sometimes
>think in Finnish. I believe the mind is incredibly deceptive, to both
>self and others. ;)
>-Jason
ooops...mistook you for someone else. Forget the "sometimes think in
English and sometimes think in Finnish" part. All the rest still apply.
-Jason
Anssi Hyytiainen
> Agree?
Yeah, it all sounds remarkably similar with idea I have about
consciousness. There's no reason to think that consciousness needs any
magic ingredient apart from "comprehension", to use your term, etc.
As far as I'm concerned, the so-called "hard problem" of consciousness
is solved.
Now someone just needs to succeed in building a system with appropriate
learning process for it to comprehend things. Then people with different
idea about "mind" will actually believe it is solved.
-Anssi
A.Isaura...@gmail.com
Interesting.
Mail me, and I WILL answer to your questions.
Don't feel lonely no more cause I do go that far.
Neo.
crois...@gmail.com
A.Isaura...@gmail.com
You suck.
You got caught in the same ego thingy thatyou so despise of the others.
=P.
Anyway got you A-OK. Interested in your theories. We are on the same
thinking scheme.
Mail me some time.
crois...@gmail.com
Anthony G. Rubino
the product of a considerable amount of serious thought. However, all
that you say in support of the position that machines are, or can be,
conscious or intelligent in a conscious sense, can also be considered as
support for the position that they are not. It all depends on a critical
difference that you acknowledge as being a result of a critical choice.
The critical difference that becomes obscured once that critical choice
is made is precisecly the fact that such a choice can be made by humans
but not by machines. Your understanding of that fact is included in the
words you use in your concluding paragraph quoted above, but it is
obscured by the inherent bias included in the critical choice.
Notice the critical requirement of motivating the machine so that its
actions will be directed towards achieving some goal. Motivation can be
considered in terms of causality, but it is a term that is usually
restricted to the willful direction of actions to achieve a goal by a
person.
Notice too that the motivation of seeking light, or hiding from it, are
human (your) motivations, not those of the machine and that you must
choose one or the other, but not both. Humans can choose one, and on a
whim, choose the other. Light can be enjoyed at one time and be abhored
at another. Humans can also have mixed feelings and motivations that
cause them to vassalate or remain immobilized. Such behaviour in
machines would render them incomprehensible unless they could be given a
free will. All the arguments against free will can be legitimately
applied to machines, but not to humans who require a will to raise those
arguments for their own purposes.
Tony, philosopher
http://www.geocities.com/trisector/
So many misconceptions, so little time.
Curt Welch
> Curt Welch wrote:
> > He talks about it under the section: "7 Learning Machines"
> >
> > He says: "Instead of trying to produce a programme to simulate the
> > adult mind, why not rather try to produce one which simulates the
> > child's? If this were then subjected to an appropriate course of
> > education one would obtain the adult brain. Presumably the child-brain
> > is something like a note-book as one buys it from the stationers.
> > Rather little mechanism, and lots of blank sheets"
> >
> > How is that anything other than a learning machine which starts off
> > with nothing but "blank sheets" and then fills it all in from learning?
>
> You are right, he does say that.
>
> Perhaps he should have made that section the central part of his
> argument. I would have.
Yeah, and maybe in his mind he did. Because in fact, the central point of
the initial part of the paper was just to try and argue the position that
it should be possible to make a machine intelligent. The last part on
learning machines was the advanced section saying something to the effect,
"If you follow and believe the logic in the first part of the paper (that
machines can be made to think like humans), then the next logical step is
to build a learning machine instead of trying to hand program all that
behavior into the machine). But his prime argument for building the
learning machine seemed to be that it would probably be easier (less code
to write). Where as my belief today (56 years in hindsight), is that
without the correct learning systems at work, it's just not intelligent.
Yeah, I look at evolution (physical and biological) and intelligence as the
same fundamental type of processes. The best way I know to describe it is
simply, directed change. This just means that any system which allows for
change which has at the same time, some effect causing the change to tend
to drift over time in some direction, will make progress over time in that
direction.
The physical property of the universe includes such a system because the
flow of energy is the source of change, and survival of structure is what
pushes the processes in a direction. Any structure which manges to
maintain it's structure and not change, even though the flow of energy is
trying to make it change, will tend to to last longer, and consume more of
the total matter of the universe, than the structures which manage not to
survive in the energy flow. This directed change of structure of mater to
more stable structures is the very essence of why the universe evolves.
Everything that happens in the universe, from the formation of galaxies, to
to the formation of what call life, to the formation of our brain, to the
creation of this Usenet message, all happened because of the processes of
directed change which controls the universe.
So, though we tend to use the word "evolution" most often to talk about
biological evolution, I see it as a fundamental process of the universe we
exist in which creates everything.
And, the brain (which this processes created), works on the same principle
of directed change. It's just more like a computer simulation of evolution
applied to the creation of machine behavior. The brain creates constant
behavior, and the learning hardware, which changes the design of the brain
as we learn, is the system which both creates constant change, and gives
that change a direction (directed change). The brain is simply running a
type of evolution program against itself in order to push it's behavior
towards a goal (maximize rewards as defined by the reward hardware in us
(pain pleasure definition created by our DNA)). So our behavior is not
just changing randomly, it's being pushed to change in a direction that
ultimately, improves our odds of survival.
DNA is also an "evolution machine" created by the the prime physical
evolution which exists in the universe. And it wasn't the first, nor the
last.
But the interesting fall out of this, is that intelligence is at it's core,
just an evolution system. Which explains why the universe looks like it
was created by something intelligent. The answer is that it was. The
answer is that biological evolution, is just the behavior of a machine
called "DNA based life". DNA (and a large gene pool of species), is when
taken as a whole, just yet another type of intelligent brain, hard at work
using it's intelligence to design new life forms. So DNA based life is in
fact, the product of intelligence.
And the rest of the universe (the planets, the hills, the rocks), were also
all designed by an intelligent processes of physical evolution. In other
words, the universe itself, because it creates a system of directed change,
is intelligent.
So, the answer to the debate between creationism and evolution, is that
they are both in fact right. It was all created by evolution, which as it
turns out, is an act of intelligent creation.
I don't think most creationists would like this answer, because the God of
evolution is not really the type of God they thought they were talking
about. But in the end, that's the truth about creation and intelligence.
> The basic structure of what I call "conscious process" needs to be of
> proper "class". The so-called "intelligence" follows open-ended learning
> mechanism that cannot understand anything in any other way but by
> "guessing" basically.
Yeah, I just call it directed change. I think that's the lowest level
fundamental process which creates intelligence. And it's present in the
entire universe.
> It is tagging semantical meanings around with no basic guideline. Other
> than "whatever meaning seems to "fit" due to all the other meanings, go
> for it." There's no explicit base. If there was, there could be no
> semantics. And so called "consciousness" follows when enough such
> semantical learning has taken place.
Semantics is an idea we tie to language. But when you look at from this
perspective of mechanical machines producing intelligent behavior, the idea
of semantics just translates to the hardware which caused a behavior. So
the meaning of a word I might write, is defined by the hardware which
created the word (i.e., my brain and body wrote the word "semantics" above
and the definition of that word in this example, is defined by whatever it
is in my brain, that caused me to write that at that time. So the
semantics of all words actually are defined by the hardware which produces
(speaks) the words.
Extending this to a wider scope, we find that if a rock rolls down a hill,
and makes a mark in the dirt as it skips down the hill, the mark has a
semantic value defined by the hardware which created it (basically the rock
and the hill). Or in other words, the mark means that a rock just skipped
past this spot.
You can extend this to talk about the meaning of all physical structure in
terms of the systems that caused it to be created.
And that brings us back to the physical evolution which is the intelligent
force creating all the structure in the universe through a simple processes
of directed change.
> > Yeah, and you might believe there is more to consciousness than either
> > I or maybe Turing believed. I suspect that he believed the same thing
> > I do.
>
> I suspect you and I and Turing all believe the "same thing", but you and
> I seem to look at the apparent structure of this mechanical process
> little differently. More about that later on in this post ->
Yeah, as I read more of your post, I started to realize we are very close
in our beliefs.
> > Ok, so the bottom line is that I do happen to believe that human
> > intelligence and consciousness is nothing more than a simple machine
> > specified by a simple algorithm. Maybe you believe there is something
> > more complex to it. Most people do.
>
> I'm not one of those people.
>
> > Well, I too have answered it. My answer is that the laws of physics
> > are intelligent.
>
> But not intentional? I know you are using the world "intelligent" in
> much broader sense than I have been in this thread.
Yes, I'm using it the sense I talked about above - directed change.
And yes, it is "intentional" in the sense that it's got direction. (the
"directed" part of directed change). It has the "intention" of changing in
some fixed direction. Though I know this is a very unconventional use of
the word intentional, I believe this in fact is the root cause of the
concept even though technical, in English, only humans are expected to have
intention.
> > I think computers are in fact already conscious. They are consciously
> > aware of me typing on the keyboard for example. They make conscious
> > decisions to print a document in response to me asking it to do that.
>
> The difference in our views is that I find it very useful to draw a
> clear line between the kind of "intelligence" you are talking about
> above, and something that we could call "conscious".
I once thought that. But then later I concluded you can't draw a clear
line because there is no real line. It's just a convention we created.
> Sure it is true that the line is completely semantical, but it is useful
> nevertheless;
Sure, most of language (maybe even all of it) has no clear lines
delineating the meaning of words, but it's still very useful to have
millions with unclear lines between them instead of one word. What's not
useful is to believe that these arbitrary lines are telling us something
real about the universe. It's like believing political land boundaries are
somehow real physical structures on the land. They are just lines drawn by
social convention. If you walk out into the field, you won't find lines in
the ground. (well, not all the time at least).
> We cannot be conscious of any actual mechanical processes
> going on in our head, the same way as a conscious machine would not be
> conscious of the actual algorithms springing its learning behaviour.
> That is why I think calling everything conscious is not very helpful.
Ah, but that's really not true. There are physical processes happening on
our brains (chemical and structural changes) which we are not consciously
aware of. That's true. But it's only because we don't have the correct
sensory hardware to sense it. The only sensory hardware we have in the
brain, is our neurons. Most people say we have no sensory hardware in the
brain, but they are wrong. All the neurons are sensory hardware. They
sense the activity of other neurons. And we are consciously aware of all
that sensing happening in our head. It's why we are able to sense the
thoughts happening in our head.
But we can't sense all the other changes happening in our head because we
don't have sensors converting all the other activity to signals which can
be processed by the brain. However, there is nothing stopping us from
creating those sensors. That's exactly what a PET scan or fMRI is. It's a
sensory device which senses brain activity we are normally unable to sense,
and feeds it into our system though other sensors.
And that gets to to point I'm making here. There is nothing happening in
our brain, which we can't sense, if we simply add some more sensors of the
right type. We can't sense it all happening at the same time, that's the
problem of trying to put a box inside itself. But you can sense any subset
of it happening do to the limits of what can be sensed, in this universe.
So we can in fact be fully conscious of any mechanical processes happening
in our head (which is also why this idea of subjective data being out of
the reach of science is just more errors in our belief systems). So you
can't use that as way to draw a line around the idea of consciousness.
> We could say that the relevant point to pick up from Searle's argument
> is that we are not aware of the mechanics of our mind. What Searle fails
> to see is that semantics are not some sort of magical intentional thing.
Well, it's not that he fails to see it. It chooses to believe that it is
some type of magic (well, that's not fair - he chooses to believe it's
different in some special way), and we simply choose to believe it's not.
Neither of us in fact has a proof that our position is correct. If my
position is correct, then we will for sure be able to build intelligent
machines - that's one fall out of my belief if the belief is correct. If
consciousness and meaning and semantics is more than just physics, then
until we figure out what that more is, we won't know what it means for
being able to built intelligent machines other than by having sex.
> Semantics are the things we are aware of.
I would argue that Semantics includes all the hardware in our brain we are
not normally aware of as well. The part of the semantics we are aware of
is just the part we are aware of.
> Things we don't really KNOW to
> be "the truth" the way that a computer "knows" 1+1 = 2. We only make
> guesses and build "ideas". Our ideas are formed by an algorithm still,
> but we should probably choose to use the word "consciousness" to refer
> to such "loose" ideas only, so not to lead people to think that we
> believe rocks have feelings and all that nonsense.
Well, I think that where this type of thought leads us is that all this
crap we talk about such as "feelings" all boils down to nothing more than a
special type of physical motion (change). So to try and argue that rocks
don't have feelings is to try and argue they don't move or change
structure.
What you can easily argue is that they don't have human feelings (which in
fact is what we are really saying when we say rocks don't have feelings).
But the real root meaning of "human feelings" is simply that rocks are
physically different than humans - duh. So to say that rocks don't act
like humans is as obvious as saying rocks don't act like water (they aren't
wet).
The important idea here is that if you believe the mind is nothing more
than a physical process of the brain, then all these mental terms like
"feelings" which have been limited to the domain of humans in common usage,
aren't in fact limited to humans by some special magic humans have
(consciousness). They are limited to humans only by social convention and
not by some law of nature. So the difference between Rock-feelings and
human-feelings, is only a mater of physical structure, and not this extra
magic which allows humans to have feelings but not rocks.
> After all, it is just as true to assert that "pure intelligence" doesn't
> exists in the universe at all. Just like pure intentionality doesn't
> exist either (due to everything actually working through evolutionary
> mechanics).
Well, as you see from what I wrote above, I believe intention and
intelligence just happens to be a fundamental physical property of our
universe. Human intelligence is one implementation of directed change but
there are an infinite number of other examples as well.
As we master the creation of machine intelligence, we will fill in the gaps
of this large continuum and man will no longer look special at all. He
will just be one tiny dot on a nearly infinite space of different types of
intelligence.
> The point is that such a thing as "semantical mechanics" is not an
> oxymoron (and HOW such a thing actually exists), even though many people
> (incl. Searle) think it is.
Right, I believe that. Others don't. But I stress, we have no proof that
this belief is a correct way to describe our universe. It's only a good
guess which, if correct, should lead us to some interesting places. If
it's a wrong guess, it will block us from seeing the real truth.
> So I really wouldn't use the word "conscious" when the computer "makes a
> decision" to print a document in respons to such a command. If it made a
> conscious decision it would need to understand semantically what the
> command means (having learned the meaning of the command), and then it
> would actually decide whether to print the document or not, depending on
> everything else it knows about the situation.
Most our decisions are not conscious by that definition. Most the stuff we
do we are not aware of (at least not aware of it for long). You can
videotape someone, and them later ask them why they did something, like tap
their fingers, and they would have no memory of ever doing it. I've had
people tell me to stop tapping a pen on the table in the middle of a
meeting and I had no idea I was even doing it. I looked down, say my hand
busy tapping the pen, and only then, did I become conscious of the fact I
was doing it in the secondary way you talk about above. Yet, I would say
my brain still made a decision to tap the pen on the table.
So the point I'm making here is that the way we use the word "conscious" is
not in fact not very well defined at all. If I (being a physical brain)
make a decision to tap my fingers on the table, but another part of my
brain is not yet conscious of the decision, or the follow on behavior of my
fingers, was the decision not an intelligent decision just because some
higher level part of the brain wasn't changed by the act? What does it
take to make an intelligent behavior which was not a conscious behavior?
If the computer gets feedback from the printer telling it that the page was
printed correctly, is that not very similar to the feedback I receive when
I turn my eyes towards by fingers and sense that they are taping? Is that
not just another example of the computer being consciously aware that it
not only asked the printer to print the document, but that the document was
in fact printed?
In the end, the word "conscious" is just a stupid word which doesn't
accurately describe much of anything. It's nearly as stupid as trying to
use Sun-Gods to explain why plants grow and then having an argument over
whether the evaporation of water was the work of the Sun God or something
else. Everyone tells me that Ra only causes plants to grow and that water
evaporates for other reasons and when I say that in fact, it's more correct
to say that Ra caused the water to evaporate (because the energy to make
plants grow and the energy to evaporate water all came from the same
source, every comes back and calls me a fool for not understanding Ra.
That's what I feel like many times when I try to argue the "true" meaning
of these English words that in fact are all messed up and have little
connection to reality like "consciousness" and "intelligence".
> > Yeah, but if that's is what intelligence is, then you find you really
> > can't draw a line between intelligent, and non-intelligent things. In
> > the end, you are forced to say that rocks are intelligent - just not as
> > intelligent as humans. Only human-like intelligence requires a
> > specific type of learning machine design.
>
> All the intelligence we see in the animal kingdom is the same type of
> intelligence as that of humans. Even the intelligence of worms. Class
> distinction between animal behaviour and plant or rock behaviour is
> useful, that's all. The learning capacity of most animals is just so
> much less that most of them never come to learn enough, to make any
> assumptions about the meaning of "existence" and "self". So I would even
> draw a separate (semantical) line between intelligence and
> consciousness, the argument being that there are species that are
> "semantically intelligent" but not necessarily self-conscious.
Yeah, I feel like I could draw that line as well. But the more times I try
to find these lines, the more I realize they just don't exist.
> Species-specific learning process (DNA) leads to -> instance-specific
> learning process (nervous systems) -> semantical understanding ->
> conscious experience
Yeah, that's how I see it as well. But when I try to define "conscious
experience" I find you can't limit it to just what humans do. When a rock
is hit by another rock, and the rock moves, and it physically changes a bit
as well (gets a mark on it), it has had a conscious experience. It is aware
of the event because it was changed by the event. In the end, that's the
root meaning of "conscious experience" or "being aware".
> > The current problem is that humans are so far separated from the next
> > most intelligent thing on the planet, that we are not left with the
> > impression that we are part of a large continuum.
>
> I was just wondering how many animals could actually learn little bit of
> human language if they had a proper throat for it. That should reveal
> people something about animals.
Yeah, I strongly suspect that one day, we will do a bit of very minor
genetic engineering and maybe a little implantation of hardware and end up
with talking mice.
> There is some simple understanding of "meaning" going on here between
> food rewards and a "voice trick" these cats perform:
> http://www.heavy.com/heavy.php?videoPath=/content/afv/flash_video/AFHV_Ta
> lkingCats (takes a while to load)
>
> The animals don't know the semantical meanings that WE have placed onto
> the "words" these animals are saying, but there is "a" meaning
> nevertheless.
And when you learn to see meaning as nothing but the physical cause of
behavior, then you see meaning everywhere. The problem is that the word
"meaning" is typically associated with language which has some magic type
of semantics which grows from the magic of consciousness. When I wave my
hand, that behavior has just as much meaning as when I wave my fingers and
type the word "meaning" on the computer. It's all just a product of the
behavior of my body. And the true meaning of both behaviors are defined by
the hardware which created the behavior.
> The "task" is still of the same class as language, or any
> understanding of meaning. Evidently cats can understand the meaning of
> the sound of a fridge being opened, or the relatively complex voice
> pattern that is their own name. Some animals seem to understand
> surprising amount of "meanings".
Yeah, how to implement the level of meaning our brain, and cat brains,
are able to create, is not trivial to figure out.
> I'm sure that gorillas could learn to utter real words with
> understanding their meaning if they had a proper throat. The same goes
> for dolphins, who have in fact a developed a kind of language and each
> have a unique name they use to address different individuals.
>
> >>>>Anyway, I doubt this to be the case, because if Turing was indeed
> >>>>having the same idea that I do, he would not have proposed such a
> >>>>thing as Turing's test at all.
> >>>Right. But now that you have proposed it, it's the right answer?
> >>
> >>I don't propose any test that is anything alike with Imitation Game.
> >
> > Didn't you say you thought the way to test intelligence was to give the
> > machine a comprehension test and see what type of answers it produced?
> > To me, this is exactly what the Turning test, and the imitation Game is
> > all about.
>
> Hmmm, could it be that you are mixing me with "Lucky"? I'm pretty sure
> I'm not him, even if he is claiming so :)
It's possible. I started responding to both of your posts around the same
time. I might have confused some of your words.
> And even if we live in the same brain, I am not responsible of his ideas
> :)
>
> >>I wouldn't if I had built a system mimicing external behaviour
> >>directly. Obviously I would if I had built a learning system. I believe
> >>my own consciousness is completely mechanical at heart. I thought I
> >>made that pretty clear.
> >
> > Ok. Some things you say seem to contradict other things you say so I'm
> > not always sure what your position is and how similar it is to mine or
> > where we might differ.
>
> Yeah, it probably seems like I am contradicting myself on a regular
> basis, but that's just because I have drawn a semantical distinction
> between mechanical processes and semantical understanding that emerges
> from a mechanical process. "Semantical mechanics."
I think if you look at it long enough, that line of yours will vanish from
the universe.
> What I mean with "semantical" is not exactly the same thing as when
> Searle is talking about semantics, since he considers it to be somehow
> fundamental. Fundamental intentionality or something like that. That's
> nonsense.
Well, I see it as nonsense. But I understand why it's just as valid for
someone else like Searle to see my position as nonsense. But the one of us
that is right will get to built the world of tomorrow, and the other will
just be forgotten as a fool. Time will be the test.
> >>Just out of curiosity, what kinds of senses are you blessing your
> >>networks with? In what form do they see the maze around them?
> >
> > Mostly very simple senses.
>
> My view of what "semantical mechanics" are, would suggest that
> sophisticated senses should be hugely beneficial for the AI. The more
> information you feed into the system, the more chances it has to draw
> all kinds of meanings of these things.
>
> For example, if I wanted to build a system that would learn to move
> around in arbitrary environment, I would think the system would be
> better off with;
> 1. proper visual information of its surroundings
> 2. orientation information of its own body (balance organ)
> 3. sense of touch (to allow it to learn to take into account the "feel"
> of if there is something underneath its feet or not)
>
> At least those three I think should be the starting point. If it just
> has a sense of whether it can move to one of four directions or not,
> there's not much meaning to be "picked up" from that.
Yeah, and many of the toy robots these days look stupid as hell because
both because they don't have enough senses and because they wouldn't know
what to do with them even if they had them (as my Aibo dog for example runs
into a wall but keeps moving it's legs as if it were still walking
forward).
But, as long as you get the data in to the system, it's really not
important what type of senses it has. Aibo for example probably has all
the senses it needs because it's got a camera it can point in different
directions. If it were in fact walking for a reason, it would have learned
to use it's vision to understand the fact that it's leg motions were not in
fact getting it closer to it's goal so it would have long ago learned to
change it's behavior in order to reach it's goal.
More senses are good only as long as you have the computing power to
processes all the data. Given a limited amount of computing power (always
the case), you don't want as much as data as you can get, you want only the
most useful data to your cause up to the limit of what you can deal with.
This is the issue of sensory overload that happens with complex
environments like jet fighter pilots that are given more data than they can
make use of and end up performing worse because of the extra sensory data
instead of better.
> Of course the good news are, that no matter how sophisticated the
> sensory systems, you don't need to worry about anything but to feed all
> the information, in whatever arbitrary form, into the system, and it
> should pick up the meaning of it. Learn how to "see", for example.
Yes, that's true. But, what it can learn to "see" in the data will always
be limited to the physical processing power of the system. In my networks
for example, they have the power to correlate current data with past data.
But how far back in time does it go? With my network, that's a function of
the topology of the network (and the the size). But by making a network
that fans out more, but is less deep, you can make it look for correlations
farther back in time, but you give up layers of abstraction. Or you can
make it deeper, but not as wide, and you get deeper layers of abstraction,
working on more current data.
So for my type of network, the number of nodes is an obvious factor in what
it can learn to "see" in the data you feed it, but also the topology of how
you choose to wire up that limited number of nodes also has a huge effect
on what it can see in the data. I suspect that any algorithm for doing
this type of correlation will always have trade offs like that which will
need to be optimized for the type of problems you expect the machine to be
solving.
Anssi Hyytiainen
> Yeah, I look at evolution (physical and biological) and intelligence as the
> same fundamental type of processes. The best way I know to describe it is
> simply, directed change.
Directed seems like a really odd word to use there in my mind. Do you
really think it is useful to consider water flowing down the hill
something that is being "directed"? Rather than something that
necessarily manifestates itself from the lower-level mechanics, such as
gravity, fluid properties and the hill properties, etc...
I mean I understand whatever is semantically "directed" is at the heart
of it all actually always mechanic, but that should be all the more
reason to see everything becoming less and less "intelligent" the
further we dig down into the mechanics of everything.
> I don't think most creationists would like this answer, because the God of
> evolution is not really the type of God they thought they were talking
> about. But in the end, that's the truth about creation and intelligence.
Well they have this agenda about "morality of God" included into their
argument about ID. Heaven and Hell and all that. They actually believe
world would be a better place if their own God with his "moral values"
is included (and such a firm belief causes only self-righteousness, and
so has been the cause of many conflicts, so that's that about the
"better place").
>>It is tagging semantical meanings around with no basic guideline. Other
>>than "whatever meaning seems to "fit" due to all the other meanings, go
>>for it." There's no explicit base. If there was, there could be no
>>semantics. And so called "consciousness" follows when enough such
>>semantical learning has taken place.
>
> Semantics is an idea we tie to language.
I would like to point out that I'm using the word "semantics" to refer
to anything at all that we can think of. Not just language. Any concept
you can be consciously aware of at all.
So the difference between you being aware of something consciously, and
some system being aware of something explicitly is that...
> Extending this to a wider scope, we find that if a rock rolls down a hill,
> and makes a mark in the dirt as it skips down the hill, the mark has a
> semantic value defined by the hardware which created it (basically the rock
> and the hill).
...the mark on the hill has no semantical value to the rock nor the
hill. It's just mark. Actually it's not even a "mark" because there's no
such thing as "mark" other than what we imagine a "mark" to mean.
The semantical meaning the mark has on us is that we just make an
assumption as to what it is when we sense it. And then we might think
someone was digging there. Or we might see the rock and think that the
rock made the mark perhaps. We don't know "what" the mark is. We don't
even know what "a mark" is.
The rock and the hill "know" what it is, but they only "know" it,
without it having any meaning apart from "truth" that there is a mark,
or a hole, a space, patch of dirt, something that we don't actually have
a "correct" word/expression for. As far as the rock and the hill are
concerned, the mark just "is", they don't make any guesses as to what
the meaning of whatever they are sensing is. Our semantical
understanding of the hole is just an expression which is not "the truth"
any more than anything else you can possibly think of.
>>The difference in our views is that I find it very useful to draw a
>>clear line between the kind of "intelligence" you are talking about
>>above, and something that we could call "conscious".
>
> I once thought that. But then later I concluded you can't draw a clear
> line because there is no real line. It's just a convention we created.
I know what you are saying, but then don't you think if you go down that
road, it becomes rather cumbersome because there is actually no clear
line between any two concepts. The lines between things are all just
conventions. We cannot see the truth of anything, we just interpetate.
We could say that everything you can think of is in reality just bunch
of quarks doing whatever they are always doing, but even that argument
is hopelessly muddled into conventions, and whatever semantical meaning
one has tagged onto things like "quarks" and "doing", and "are" and "is"
and so on. These are all concepts that only exist in our imagination.
That we have an imagination, is also because of bunch of quarks doing
whatever they are always doing, but since they are locked into such
process as they are in our brain, it is completely justified to assert
that we have so-called imagination and rocks don't. It is justified even
though no one can point out the "true" line between rocks and our brain.
If the impossibility to draw the line alone made this assertion false,
then we could not assert anything at all. We could not claim anything to
be a table or a chair, since there is no clear line between the two.
But you have to make assertions of things in order to be able to think
of anything consciously.
>>We cannot be conscious of any actual mechanical processes
>>going on in our head, the same way as a conscious machine would not be
>>conscious of the actual algorithms springing its learning behaviour.
>>That is why I think calling everything conscious is not very helpful.
>
> Ah, but that's really not true. There are physical processes happening on
> our brains (chemical and structural changes) which we are not consciously
> aware of. That's true. But it's only because we don't have the correct
> sensory hardware to sense it. The only sensory hardware we have in the
> brain, is our neurons. Most people say we have no sensory hardware in the
> brain, but they are wrong. All the neurons are sensory hardware. They
> sense the activity of other neurons. And we are consciously aware of all
> that sensing happening in our head. It's why we are able to sense the
> thoughts happening in our head.
>
> But we can't sense all the other changes happening in our head because we
> don't have sensors converting all the other activity to signals which can
> be processed by the brain. However, there is nothing stopping us from
> creating those sensors. That's exactly what a PET scan or fMRI is. It's a
> sensory device which senses brain activity we are normally unable to sense,
> and feeds it into our system though other sensors.
These signals still have to be expressed in some form to our conscious
mind before they are semantically sensical. What is the "true form" to
express electrical signals in? Any form of expression one can think of
is just something they'd think to be useful in one way or another. But
no form of expression is "the truth" or "true form" of these signals.
For a single neuron, an incoming signal is "truth" in a way we can never
consciously "understand". We can only imagine something about it, and
this very imagination is just one form of expressing this signal and the
neuron's reaction to it. Choosing the form of expression is a way to tag
meaning on something.
There's a similar rather interesting muddle of conventions in physics,
when we are trying to find out the truth of what an electron is, for
example. One classical way to express is, is to think of a particle
orbiting the nucleus like earth orbits the sun, and that has pretty much
stuck to us because someone made that sort of analogy when first
describing the properties of the electron, and how it can move to outer
or inner orbits of the atom (although he made it clear he was not
attempting to describe the "true" nature of electron, just its functions
through analogy).
But then electron also has wave properties, and it doesn't "move" from
one orbital to another, it just disappears from one orbital and
reappears to another. So it is not a particle orbiting the nucleus, it
is something else we haven't been able to express yet properly.
"Particle and a wave" is about the best we've come up with so far.
I'm mentioning this because there is also some rather annoying
interpetations taking place here. Such as, that the electron needs to be
observed by "a conscious mind" before it's wave-function collapses and
it is found from some position (assertion being that it is not enough
for a "non-conscious" machine to detect the electron, for example). That
is the amount of desperation we have in clinging to the concepts we have
imagined to be "true".
>>So I really wouldn't use the word "conscious" when the computer "makes a
>>decision" to print a document in respons to such a command. If it made a
>>conscious decision it would need to understand semantically what the
>>command means (having learned the meaning of the command), and then it
>>would actually decide whether to print the document or not, depending on
>>everything else it knows about the situation.
>
> Most our decisions are not conscious by that definition.
Let's say, things we "can be" conscious of. There are things we are
paying attention to, and then there are things we "can think of".
>>Species-specific learning process (DNA) leads to -> instance-specific
>>learning process (nervous systems) -> semantical understanding ->
>>conscious experience
>
> Yeah, that's how I see it as well. But when I try to define "conscious
> experience" I find you can't limit it to just what humans do. When a rock
> is hit by another rock, and the rock moves, and it physically changes a bit
> as well (gets a mark on it), it has had a conscious experience.
What about making class distinction between systems that can predict
things like "that falling rock is going to hit the ground soon", and
things that can only react to things when they happen? (After all, that
is the same class distinction as between things that can think and
things that can't. And things that can, they have "imagination" by
definition)
-Anssi
Curt Welch
> Curt Welch wrote:
> > Yeah, I look at evolution (physical and biological) and intelligence as
> > the same fundamental type of processes. The best way I know to
> > describe it is simply, directed change.
>
> Directed seems like a really odd word to use there in my mind. Do you
> really think it is useful to consider water flowing down the hill
> something that is being "directed"? Rather than something that
> necessarily manifestates itself from the lower-level mechanics, such as
> gravity, fluid properties and the hill properties, etc...
It's the same either way as I see it. The shape of the hill is directing
the path of the water as it flows down the hill. The concept of directed
change that I'm trying put forth here is simply that there is some
combination of low level physical circumstances which is causing the water
to move/change in a specific direction instead of just moving randomly.
> I mean I understand whatever is semantically "directed" is at the heart
> of it all actually always mechanic, but that should be all the more
> reason to see everything becoming less and less "intelligent" the
> further we dig down into the mechanics of everything.
Yeah, I think that's true. There's very little intelligence in a atom.
The point isn't that how much intelligence there is down there, the point
is that it's the same basic force which drives all evolution and all
intelligence in this universe. It's not some magic force. It's just the
fact that the nature of matter in this universe is such that 1) it changes
structure over time and 2) the very lowest level nature of how it changes
allows for the system to direct the change - some structures are able to
last longer in the face of change than others and that gives it direction.
The direction is from structures which change structure easily (plasma) to
structures which actively defend their existence (planets, rocks, humans).
> > I don't think most creationists would like this answer, because the God
> > of evolution is not really the type of God they thought they were
> > talking about. But in the end, that's the truth about creation and
> > intelligence.
>
> Well they have this agenda about "morality of God" included into their
> argument about ID. Heaven and Hell and all that. They actually believe
> world would be a better place if their own God with his "moral values"
> is included (and such a firm belief causes only self-righteousness, and
> so has been the cause of many conflicts, so that's that about the
> "better place").
Yeah, but in fact, this God with morality does exist. Morality is the
direction created by the system of directed change. In the human brain,
our morality is created and defined by the hardware which defines pain and
pleasure. That's the hardware which gives our learning brain direction to
know what to learn and what to forget and at the highest levels, this low
level mechanical direction created by the hardware is what comes out as our
debates of morality (i.e., what should we do and not do - i.e. which of two
behaviors is better). The concept of heaven is the idea of pure pleasure
and the concept of hell is the idea of pure pain and we are forced to live
in the middle constantly trying to balance the two as we attempt to
maximize our reward (get to heaven). This is simply what we are. We are
machines which are built to spend their life trying to maximize total
reward (do as much good in our life as we can before we die).
Below the level of our brain is the physical behavior of the universe which
gives everything direction (towards forming structures that last longer -
aka survival) which is the prime morality of the "God" which created us.
So what these religious people know is true, is in fact basically true. So
if we look at evolution and interpret it to mean that there is no God and
there is no morality, and that nothing cares what happens, we are wrong.
The universe does in effect care which direction the universe moves in.
In other words, I don't think there really is a science vs religion choice
we have to make here. I believe that all of religion, and faith, and
belief, can be explained and understated in scientific terms. If we can
explain the mind in terms of physical science (which I believe is possible
in this universe since all indication is that the mind is just a physical
manifestation of the brain), then we would be able to explain everything
except for the one question which will always be unanswered - why are we
here.
The religious right tend to push agendas that are very different from the
scientific community, but you can't use science to prove their agenda is
wrong. In fact, much of what I see about science tends to only support
many of their beliefs. The prime difference is a personal choice people
tend to have made between putting their faith in reason (it's not right
unless I can logically explain it) or putting their faith in faith (if it
feels right, it is right).
It's not any better to use science to push your agenda than to use faith to
push your agenda. In the end, you are doing the same thing either way -
pushing your own personal agenda on the rest of the universe which already
has it's own agenda for all of us.
> >>It is tagging semantical meanings around with no basic guideline. Other
> >>than "whatever meaning seems to "fit" due to all the other meanings, go
> >>for it." There's no explicit base. If there was, there could be no
> >>semantics. And so called "consciousness" follows when enough such
> >>semantical learning has taken place.
> >
> > Semantics is an idea we tie to language.
>
> I would like to point out that I'm using the word "semantics" to refer
> to anything at all that we can think of. Not just language. Any concept
> you can be consciously aware of at all.
Yeah, I guess that's typical of how we all normally use the word.
> So the difference between you being aware of something consciously, and
> some system being aware of something explicitly is that...
>
> > Extending this to a wider scope, we find that if a rock rolls down a
> > hill, and makes a mark in the dirt as it skips down the hill, the mark
> > has a semantic value defined by the hardware which created it
> > (basically the rock and the hill).
>
> ...the mark on the hill has no semantical value to the rock nor the
> hill. It's just mark. Actually it's not even a "mark" because there's no
> such thing as "mark" other than what we imagine a "mark" to mean.
>
> The semantical meaning the mark has on us is that we just make an
> assumption as to what it is when we sense it. And then we might think
> someone was digging there. Or we might see the rock and think that the
> rock made the mark perhaps. We don't know "what" the mark is. We don't
> even know what "a mark" is.
Right, there seems to be two sides to the story of semantics but there's
not. In language, there is the meaning of the word when it's generated.
That is, the meaning of the word as defined by the person who spoke the
word. Then there is the effect that word has on others, which is the
meaning as perceived by the listener. So for the mark created by the rock,
there is the meaning as defined by the rock who created the mark, and there
is the meaning as understood by us, when we sense the existence of the
rock. And like you show, the meaning we understand when we see the mark,
can be very different. We might in fact not understand what the rock was
trying to tell us (that it's was a rock rolling down a hill making a mark
in the dirt).
But this is all the same type of meaning in the end, because the meaning of
the mark as it was created, is ultimately defined by the hardware which
created the mark. And the meaning of the mark, when we sensed it, was
ultimately defined by the hardware which sensed it. And the act of sensing
is in fact always an act of communicating (receiving one signal and sending
out another). So the meaning we assign to the sensed signal is defined by
the signal we generated in response to it. The simple act of perception,
is implemented internally, as hardware which receives one messages, and
then sends out another saying "I just heard X". And the "I just heard X"
message, is the meaning which was assigned to the sensation, and the
meaning of this meaning, is defined by the operation of the hardware.
So, in all causes, it gets back to "meaning" just being defined by the
physical nature of the hardware which is involved in the physical
interaction.
> The rock and the hill "know" what it is, but they only "know" it,
> without it having any meaning apart from "truth" that there is a mark,
> or a hole, a space, patch of dirt, something that we don't actually have
> a "correct" word/expression for.
Yeah, isn't it fun trying to talk about these things in a language that
just doesn't have words which correctly explain the ideas.
> As far as the rock and the hill are
> concerned, the mark just "is", they don't make any guesses as to what
> the meaning of whatever they are sensing is. Our semantical
> understanding of the hole is just an expression which is not "the truth"
> any more than anything else you can possibly think of.
Right. Our understanding is far more complex than the rocks because we
don't have one device communicating some opinion we have billions of them
in our head all communicating different ideas. Each neuron has different
connections to other neurons and as a result, they will all "talk" at
different times, in response to different effects - which means they all
have different meanings and "understanding". Our human understanding is
the sum total of the understanding of all these billions of neurons. A
rock understands only the information that passes though it - which is more
like the understanding of a single neuron (probably even less). So it's
not that rocks don't understand things, or don't communicate it's
understanding to others, it's just that a human brain is at least 100
billion times smarter than a rock and has 100 billions times greater
understanding of it's environment. We have that much more semantic
understanding power than a rock.
> >>The difference in our views is that I find it very useful to draw a
> >>clear line between the kind of "intelligence" you are talking about
> >>above, and something that we could call "conscious".
> >
> > I once thought that. But then later I concluded you can't draw a clear
> > line because there is no real line. It's just a convention we created.
>
> I know what you are saying, but then don't you think if you go down that
> road, it becomes rather cumbersome because there is actually no clear
> line between any two concepts. The lines between things are all just
> conventions. We cannot see the truth of anything, we just interpetate.
We just have to use finer detail to create the lines. The problem is that
the concept of consciousness is not the correct language to use to define
the lines. If for example, we know how to build something with human
intelligence, and if it was some type of neural network, then we could talk
precisely about how many nodes a given device had and what type of network
topology it used and what type of sensors it had and exactly how many (a
300x200 pixel eye for example). These are the details you use to describe
the level of consciousness.
We don't for example have much problem talking about, or understanding,
that there are many different types of computers in the world and we can
talk about all the details of their construction and understand their
differences by comparing details (one has more memory, more less disks,
runs faster). But if you were to try and find where the line was between
computer, and non computer, you could not really find it. How many logic
gates do you have to hook together before they become a computer? Is a
mechanical calculator a computer? What features do you have to add to a
mechanical calculator before it becomes a computer? This is where you
would have real problems trying to draw the line - in the end, you find
there is no line, there is just a lot of grey area where you can, if you
want, just pick some answer at random to define the line - but that the
line didn't exist as a fundamental aspect of the nature of matter. It was
created by mere social agreement on how we should use the word "computer".
We have general social agreement on the correct use of the word
consciousness because we mostly limit it's use to describing humans. There
isn't anything else in the universe which comes close so we use the human /
not human line to define the correct social use of the word consciousness.
But, once we look at the problem of making machines conscious, the old
social conventions won't keep working, and we will have to use new concepts
of finer resolution to tell the difference between different types of
consciousness.
And until we actually know how to create machine consciousness, we won't
know what those words will be - so it's hard to talk about it now.
> We could say that everything you can think of is in reality just bunch
> of quarks doing whatever they are always doing, but even that argument
> is hopelessly muddled into conventions, and whatever semantical meaning
> one has tagged onto things like "quarks" and "doing", and "are" and "is"
> and so on. These are all concepts that only exist in our imagination.
Well, yes and no. We have tools to verity our imagination when it comes to
the nature of the physical universe and tools to synchronize our
imagination with the imagination of others. There will always be confusion
about the gray areas, but if you create and use concepts of fine enough
detail, then you can minimize that confusion to just about any degree you
want.
> That we have an imagination, is also because of bunch of quarks doing
> whatever they are always doing, but since they are locked into such
> process as they are in our brain, it is completely justified to assert
> that we have so-called imagination and rocks don't. It is justified even
> though no one can point out the "true" line between rocks and our brain.
Yes, but without knowing the true cause of our imagination, the decision to
declare that it's something humans have and rocks don't makes it's a purely
social convention and not useful information about the nature of the
universe.
And based on what I believe consciousness actually is, rocks and computers
do already have it. It's only a question of what type of consciousness
they have vs what type of consciousness humans have.
> If the impossibility to draw the line alone made this assertion false,
> then we could not assert anything at all. We could not claim anything to
> be a table or a chair, since there is no clear line between the two.
>
> But you have to make assertions of things in order to be able to think
> of anything consciously.
Yes, you must draw arbitrary lines in order to classify things as different
so that we can recognize them when we see them again. We have no problem
saying humans are conscious and rocks aren't because the difference is so
huge and so easy to identify. But to create consciousness, we must
understand what creates it and not just how to classify rocks and humans.
And my current best understanding of what consciousness is tells me it's a
continuum which rocks are a part of.
So the issue here isn't just social convention, it's an issue of how the
universe works. I used to see intelligence and consciousness as something
that would only result from the correct type of machine/processes, as you
seem to believe now. But I now believe it is in fact a nearly infinite
continuum of variation that includes humans near one end and rocks near the
other end.
There is no point that we can create a single machine and say that we have
crossed the line from unconscious to conscious machine. All we can do is
create machines closer and closer in ability to humans. The closer we get,
the more obvious it will become that humans are just a part of a huge
continuum of machine designs and not something unique and special in the
world.
From the other end of the spectrum, in the machine world, we have the idea
of machine learning. That too is such a vague concept as to have almost no
meaning. All machines in fact learn to some degree because they all change
as a result of interaction with other things and the fundamental meaning of
"learning" is nothing more than "change due to interaction".
So, just saying that the solution to AI is a learning machine (like I like
to do) tells us almost nothing more than saying the solution to AI is
building a conscious machine. The devil is in the details, and the details
can vary to such an extent as to create an infinite continuum of learning
machines from something similar to rocks to something similar to humans.
The real trick to solving AI is creating a learning machine with powers
very similar to humans, not to just create a learning machine, which is
trivial to do.
I guess on one level I grasp what you are saying. But you have to be
careful because we are in fact that neuron. We are not separate from it.
(even though humans have a long history of talking as if we were separate
from our bodies.
I am a neuron (plus a few billion other neurons plus some arms and some
legs etc.).
So, to say I can't understand the neuron when I am the neuron is a bit odd
to say. My understanding and the neuron's understanding is one and the
same understanding.
> We can only imagine something about it, and
> this very imagination is just one form of expressing this signal and the
> neuron's reaction to it. Choosing the form of expression is a way to tag
> meaning on something.
Yes, but that's one collection of neurons in my head trying to understand
some other neuron in my head. It's true that neuron B can't fully
understand the meaning of neuron C. But neuron B is my understanding of
neuron C. And neuron C is my understanding of whatever it is that neuron C
understands. And since I am in fact both neuron B and C combined, I fully
understand it all.
This gets down to an odd issue of reference when trying to talk about what
understanding is.
If rock J hits rock K, then rock K moves. And that movement is rock Ks
"understanding" of rock J. The movement in rock K in no way represents a
full and complete understanding of rock J, so rock K has only a partial
understanding of rock J. But rock K's understanding is not just some
imagination of rock J. It's real and complete. It is what it is. Rock K
moved because of some external force and that movement is it's
understanding of the external force.
> There's a similar rather interesting muddle of conventions in physics,
> when we are trying to find out the truth of what an electron is, for
> example. One classical way to express is, is to think of a particle
> orbiting the nucleus like earth orbits the sun, and that has pretty much
> stuck to us because someone made that sort of analogy when first
> describing the properties of the electron, and how it can move to outer
> or inner orbits of the atom (although he made it clear he was not
> attempting to describe the "true" nature of electron, just its functions
> through analogy).
>
> But then electron also has wave properties, and it doesn't "move" from
> one orbital to another, it just disappears from one orbital and
> reappears to another. So it is not a particle orbiting the nucleus, it
> is something else we haven't been able to express yet properly.
> "Particle and a wave" is about the best we've come up with so far.
Yes, we describe and understand everything by listing their properties. If
we are lucky, we find a ways to simplify the list by using analogy.
Sometimes the analogies are taken too far, and produce unexpected errors in
the simplification. (electrons really aren't just particles or waves).
> I'm mentioning this because there is also some rather annoying
> interpetations taking place here. Such as, that the electron needs to be
> observed by "a conscious mind" before it's wave-function collapses and
> it is found from some position (assertion being that it is not enough
> for a "non-conscious" machine to detect the electron, for example). That
> is the amount of desperation we have in clinging to the concepts we have
> imagined to be "true".
Yeah, not having a clear understanding of what the mind is gets in the way
of science since the mind is the prime tool we use for doing science.
> >>So I really wouldn't use the word "conscious" when the computer "makes
> >>a decision" to print a document in respons to such a command. If it
> >>made a conscious decision it would need to understand semantically what
> >>the command means (having learned the meaning of the command), and then
> >>it would actually decide whether to print the document or not,
> >>depending on everything else it knows about the situation.
> >
> > Most our decisions are not conscious by that definition.
>
> Let's say, things we "can be" conscious of. There are things we are
> paying attention to, and then there are things we "can think of".
Yeah, that works.
> >>Species-specific learning process (DNA) leads to -> instance-specific
> >>learning process (nervous systems) -> semantical understanding ->
> >>conscious experience
> >
> > Yeah, that's how I see it as well. But when I try to define "conscious
> > experience" I find you can't limit it to just what humans do. When a
> > rock is hit by another rock, and the rock moves, and it physically
> > changes a bit as well (gets a mark on it), it has had a conscious
> > experience.
>
> What about making class distinction between systems that can predict
> things like "that falling rock is going to hit the ground soon", and
> things that can only react to things when they happen? (After all, that
> is the same class distinction as between things that can think and
> things that can't. And things that can, they have "imagination" by
> definition)
But how is a prediction not really just a reaction? When you look at what
happens when we make predictions all you find at the lowest level is
nothing but reactions and more reactions.
If the lights of a car trigger an electric eye that causes the garage door
to open before the car gets there which allows the car to enter without
hitting the garage door, did the electric eye predict the arrival of the
car? We wouldn't normally talk about the operation of such as machine as
making predictions.
But what happens in us when we make predictions works exactly like that.
It's just a chain of physical reactions. It's all just a chain of physical
reactions at the lowest level. So how would you define which physical
reactions were "predictions" and which were not? Off the top of my head, I
can't see how you would do that.
I think the brain is a reinforcement learning machine which attempts to
assign value to behavior based on it's estimation of total future rewards
each behavior is expected to yield. So our brain act as prediction
machines. But you can look at the operation of a machine from many
different perspectives, (all of them valid). It's just as easy to define
the function in terms which don't use the idea of prediction. You can just
say that a reward increased the probability of a behavior being used in the
future and not talk about predictions at all.
The concept of prediction isn't a very good concept for accurately
describing machine behavior - which is why we don't tend to use it all that
much when we describe the design and operation of a machine.
Anssi Hyytiainen
> Right, there seems to be two sides to the story of semantics but there's
> not. In language, there is the meaning of the word when it's generated.
> That is, the meaning of the word as defined by the person who spoke the
> word. Then there is the effect that word has on others, which is the
> meaning as perceived by the listener.
Actually even when the word is being generated, its meaning is "loose"
because it relies on whatever other meanings the generating mind has
placed on other things. Different meanings support each others in the
worldview, but they also have an effect on each others, should the
meaning of things change because of learning something new.
> And the meaning of the mark, when we sensed it, was
> ultimately defined by the hardware which sensed it. And the act of sensing
> is in fact always an act of communicating (receiving one signal and sending
> out another). So the meaning we assign to the sensed signal is defined by
> the signal we generated in response to it.
...which is defined by everything the machine has learned. The response
is completely subject to its earlier experiences and what it has picked
up from them due to things it has learned even earlier.
It's pretty important to keep that part intact when describing the
reaction our brain has to the signals from the sensory systems.
> The simple act of perception,
> is implemented internally, as hardware which receives one messages, and
> then sends out another saying "I just heard X". And the "I just heard X"
> message, is the meaning which was assigned to the sensation, and the
> meaning of this meaning, is defined by the operation of the hardware.
...as it operates at the state it is in due to prior learning, etc.
> The problem is that
> the concept of consciousness is not the correct language to use to define
> the lines.
Yeah, but you know, what is... Argumenting about anything can be
problematic due to semantics, and argumenting about consciousness is
twice that fun because then you are basically argumenting ABOUT semantics...
>>We could say that everything you can think of is in reality just bunch
>>of quarks doing whatever they are always doing, but even that argument
>>is hopelessly muddled into conventions, and whatever semantical meaning
>>one has tagged onto things like "quarks" and "doing", and "are" and "is"
>>and so on. These are all concepts that only exist in our imagination.
>
> Well, yes and no. We have tools to verity our imagination when it comes to
> the nature of the physical universe and tools to synchronize our
> imagination with the imagination of others.
But do we really? We can only get closer to the truth, or at least think
that we are getting closer, but how could we ever hold the true nature
of anything inside our imagination? Which is just an artificial
expression of something in nature.
> I guess on one level I grasp what you are saying. But you have to be
> careful because we are in fact that neuron. We are not separate from it.
> (even though humans have a long history of talking as if we were separate
> from our bodies.
>
> I am a neuron (plus a few billion other neurons plus some arms and some
> legs etc.).
>
> So, to say I can't understand the neuron when I am the neuron is a bit odd
> to say.
Well I wouldn't say it is odd to say at all, because we should expect
your conscious understanding to rise from such a process that is
basically expressing "things you observe" in some imaginary form anyway.
The fact that this imaginary form manifestates itself from the processes
that the neurons are causing doesn't mean the imaginary form IS the
neurons, even if you are imagining a neuron itself.
>>What about making class distinction between systems that can predict
>>things like "that falling rock is going to hit the ground soon", and
>>things that can only react to things when they happen? (After all, that
>>is the same class distinction as between things that can think and
>>things that can't. And things that can, they have "imagination" by
>>definition)
>
> But how is a prediction not really just a reaction? When you look at what
> happens when we make predictions all you find at the lowest level is
> nothing but reactions and more reactions.
Sure it's a reaction, but it is a reaction based on your GUESS as to
what is going to happen in the future. This "guess" is a manifestation
of a chain of reactions (which is subject to what you have learned),
sure, but it is still something you are just imagining MIGHT happen, and
you act accordingly. Such a behaviour brings about creativity, among
other things. Creating and inventing things is basically about guessing
how things might unfold if you did this or that. Building your networks
is about guessing how things might unfold if you did something.
So if you accept the above definition of "prediction", then the electric
cat eye reacting to the garage door is not prediction, it's just
straightforward reaction to the light. For it to be able to really
"guess" what how some novel situation might unfold, it would need to
have some assumptions of the meanings of things it is observing in the
situation. Like sounds and lights. Perhaps when it saw just one
headlight, and heard no engine noises, it might guess it means there is
a bicycle coming and it doesn't need to open the door, etc.
Or perhaps it understood the "meaning" of burglars, or more to the
point, the meaning of commando masks and pick up "suspicious behaviour"
of those who are arriving to mean it shouldn't open itself. And so on.
Those all require it to build a worldview of semantical nature.
-Anssi
Curt Welch
> Curt Welch wrote:
> > Right, there seems to be two sides to the story of semantics but
> > there's not. In language, there is the meaning of the word when it's
> > generated. That is, the meaning of the word as defined by the person
> > who spoke the word. Then there is the effect that word has on others,
> > which is the meaning as perceived by the listener.
>
> Actually even when the word is being generated, its meaning is "loose"
> because it relies on whatever other meanings the generating mind has
> placed on other things.
Well, maybe. In this odd way I'm talking about meaning here, all we are
doing is tracing energy flow backwards in time. The meaning of the word
which was generated is defined by the physical characteristic of the device
which generated the word (an energy flow), but the last device in the chain
was given signals from the devices before it, which have their own
definition of meaning defined by how energy flows through them. If you try
to trace it backwards very far you very quickly end up forming a tree which
consumes the entire universe (at the rate of the speed of light expanding
outwards).
Whether you call this "loose" is an issue I guess of whether you see
existence itself, and the interaction of matter in the universe as being
"loose".
> Different meanings support each others in the
> worldview, but they also have an effect on each others, should the
> meaning of things change because of learning something new.
Yes, the meaning of a word is not constant. It can change every time it is
spoken. At the atomic level of looking at this, it might be correct that
to say that it must change every time. But that's just the complexity of
what happens in this universe at the atomic level. That doesn't stop us
from building NAND gates that have a very well defined meaning which is
able to stay very stable for a very long time.
> > And the meaning of the mark, when we sensed it, was
> > ultimately defined by the hardware which sensed it. And the act of
> > sensing is in fact always an act of communicating (receiving one signal
> > and sending out another). So the meaning we assign to the sensed
> > signal is defined by the signal we generated in response to it.
>
> ...which is defined by everything the machine has learned. The response
> is completely subject to its earlier experiences and what it has picked
> up from them due to things it has learned even earlier.
Yes, like I said, the full current meaning if you keep going back in time
represents information from the entire universe.
> It's pretty important to keep that part intact when describing the
> reaction our brain has to the signals from the sensory systems.
Oh yeah, I fully understand that. And it's not just going back to birth.
The concept extends back forever. In us, you see how our evolutionary
history controls our behavior (nature) just as much our senses extending
back to birth (nurture). It's all stuff our DNA has sensed over its life
time which has led to its current behavior of building human bodies.
> > The simple act of perception,
> > is implemented internally, as hardware which receives one messages, and
> > then sends out another saying "I just heard X". And the "I just heard
> > X" message, is the meaning which was assigned to the sensation, and the
> > meaning of this meaning, is defined by the operation of the hardware.
>
> ...as it operates at the state it is in due to prior learning, etc.
>
> > The problem is that
> > the concept of consciousness is not the correct language to use to
> > define the lines.
>
> Yeah, but you know, what is... Argumenting about anything can be
> problematic due to semantics, and argumenting about consciousness is
> twice that fun because then you are basically argumenting ABOUT
> semantics...
Yeah, I like to joke that philosophy is like trying to use a hammer to
hammer itself with. When you use language to talk about language, you get
a dangerous self referencing loop which allows you to throw about any idea
you want into the loop and make it sound valid.
> >>We could say that everything you can think of is in reality just bunch
> >>of quarks doing whatever they are always doing, but even that argument
> >>is hopelessly muddled into conventions, and whatever semantical meaning
> >>one has tagged onto things like "quarks" and "doing", and "are" and
> >>"is" and so on. These are all concepts that only exist in our
> >>imagination.
> >
> > Well, yes and no. We have tools to verity our imagination when it
> > comes to the nature of the physical universe and tools to synchronize
> > our imagination with the imagination of others.
>
> But do we really? We can only get closer to the truth, or at least think
> that we are getting closer, but how could we ever hold the true nature
> of anything inside our imagination? Which is just an artificial
> expression of something in nature.
Because what we hold in our imagination is as true as it gets for us.
Nothing can be more true for us. If there is something more complex, or
more complete, happening outside our imagination, but we can't sense it,
and we can't know it, then it will never be our truth. Our truth is not an
artificial expression of something in nature. It is the only nature there
is for us. It is the only universe there is for us. The limits of our
universe are the limits of our understanding.
> > I guess on one level I grasp what you are saying. But you have to be
> > careful because we are in fact that neuron. We are not separate from
> > it. (even though humans have a long history of talking as if we were
> > separate from our bodies.
> >
> > I am a neuron (plus a few billion other neurons plus some arms and some
> > legs etc.).
> >
> > So, to say I can't understand the neuron when I am the neuron is a bit
> > odd to say.
>
> Well I wouldn't say it is odd to say at all, because we should expect
> your conscious understanding to rise from such a process that is
> basically expressing "things you observe" in some imaginary form anyway.
The things we know do not exist in an imaginary form. They exist in the
form they exist in and that's all that exists for us. To me, I feel as if
I'm sitting in my office typing on the computer. There's all this stuff
around me which I see as being not part of my body. But in fact, it's all
part of my body. What I'm "seeing" is not really a computer sitting in
front of me. It's just a bunch of neurons firing. It's as if I'm locked
inside my head, watching pictures of the word around me. We never get to
see the real world. We are forced to only understand that which is sent to
us, instead of being able to go out and get it.
So, one neuron in my head, only understands what it understands. It
understands what it can sense, which is the neurons it's attached to. All
my understanding of the office around me comes from the sum total of all
these neurons locked in my head, only able to sense other neurons also
locked in my head.
So though it feels like I'm sensing all this stuff outside my head, it's
all an illusion. All I can sense is the stuff happening in my head.
And there's nothing imaginary about the behavior of all these physical
neurons. It's all very real physical changes happening to the neurons
which represents what each of them knows. What happens to them, is what
they know, and what they know, is all we know, because all we are, is this
collection of neurons trapped in our head.
> The fact that this imaginary form manifestates itself from the processes
> that the neurons are causing doesn't mean the imaginary form IS the
> neurons, even if you are imagining a neuron itself.
It does actually. That's what happens when you realize there is no mind
body split and fix all the problems in our language that is built on the
assumption that there is a mind body split. The brain does not represent
some imaginary form of the world outside for us as much as it _IS_ the
world to us. The internal behavior of the brain is our world. It's the
only world we can ever know.
> >>What about making class distinction between systems that can predict
> >>things like "that falling rock is going to hit the ground soon", and
> >>things that can only react to things when they happen? (After all, that
> >>is the same class distinction as between things that can think and
> >>things that can't. And things that can, they have "imagination" by
> >>definition)
> >
> > But how is a prediction not really just a reaction? When you look at
> > what happens when we make predictions all you find at the lowest level
> > is nothing but reactions and more reactions.
>
> Sure it's a reaction, but it is a reaction based on your GUESS as to
> what is going to happen in the future.
We sure we can (and do) talk that way. But as you said above, our reaction
is really better seen as nothing but a sum of our total history. This is
the trick that Behaviorism learned. When trying to understand how the
machine which is a human works, it's much cleaner to talk about our current
behavior based on our past experience than to use mentalist terms like
"prediction".
> This "guess" is a manifestation
> of a chain of reactions (which is subject to what you have learned),
> sure, but it is still something you are just imagining MIGHT happen, and
> you act accordingly. Such a behaviour brings about creativity, among
> other things. Creating and inventing things is basically about guessing
> how things might unfold if you did this or that. Building your networks
> is about guessing how things might unfold if you did something.
Well, this is tricky language stuff you get into that's the heart of
Behaviorism. You can talk like you do above, but it's just better if you
don't in terms of trying to explain what the machine is actually doing.
There is no correct way to manifest a "prediction" in hardware. There are
many different ways to do it. But there is one clear and correct ways to
manifest a reaction. When I write computer software, I can think of what
it is doing in terms of predictions (I predict which block in the buffer
cache is least likely to be used next and free up that space for the new
block), but the code I actually write ends up being a list of fixed and
well define reactions, not predictions. There are no low level computer
instruction primitives that say "predict the expected value of x".
So though it's not wrong to think about the operation of a machine in terms
of it's ability to predict, you have to build the machine in terms of all
the reactions you want it to create. So if you can just remove all your
talk about how it's predicting the future from your language that describes
the operation of the machine, then translating the description into
hardware becomes much simpler.
> So if you accept the above definition of "prediction", then the electric
> cat eye reacting to the garage door is not prediction, it's just
> straightforward reaction to the light. For it to be able to really
> "guess" what how some novel situation might unfold, it would need to
> have some assumptions of the meanings of things it is observing in the
> situation. Like sounds and lights. Perhaps when it saw just one
> headlight, and heard no engine noises, it might guess it means there is
> a bicycle coming and it doesn't need to open the door, etc.
>
> Or perhaps it understood the "meaning" of burglars, or more to the
> point, the meaning of commando masks and pick up "suspicious behaviour"
> of those who are arriving to mean it shouldn't open itself. And so on.
> Those all require it to build a worldview of semantical nature.
I'll respond to that later....
> -Anssi
Curt Welch
> Anssi Hyytiainen <ans...@nic.fi.ANTISPAM> wrote:
> > Sure it's a reaction, but it is a reaction based on your GUESS as to
> > what is going to happen in the future. This "guess" is a manifestation
> > of a chain of reactions (which is subject to what you have learned),
> > sure, but it is still something you are just imagining MIGHT happen,
> > and you act accordingly. Such a behaviour brings about creativity,
> > among other things. Creating and inventing things is basically about
> > guessing how things might unfold if you did this or that. Building your
> > networks is about guessing how things might unfold if you did
> > something.
[snip my previous comments]
> > So if you accept the above definition of "prediction", then the
> > electric cat eye reacting to the garage door is not prediction, it's
> > just straightforward reaction to the light.
And this is just the problem that type of talk gets you into.
The electric eye example is as much a prediction as what the brain does but
it doesn't seem that way because we were taught to talk about human
behavior using intentional words like "prediction" and "knowledge" and
"meaning" but we were taught to talk about the operation of a machine using
extensional words like "reactions". So since we describe human behavior
using such a different language than the language we use to describe
machine behavior, it ends up leaving us with the obvious conclusion that
the two processes are substantially different when they aren't.
If the electric eye sends an electrical signal to the motor, the meaning of
that signal can be said to be the machines internal prediction that a car
is coming. The fact that the motor then responds to that prediction by
raising the garage door can be said to be the motors reaction to the
prediction. The motor wanted to make sure the garage door was out of the
way of the car because the motor was predicting that the garage door would
be damaged if it didn't raise the door.
Now, all the above talk about the electric eye predicting the car and the
motor intentionally raising the door to get it out of the way of the car
because it was predicting damage if it didn't move the door sounds absurd
to most people. But in fact, it's the exact same absurd talk we use to
describe humans which aren't doing anything different than the electric eye
and the motor when we do the same sort of thing.
It's the past experience of the motor and the electric eye that transformed
them into this configuration what caused them to act like they act. Once
the motor failed to raise the door and a truck banged into it and knocked
the door off it's tracks. Quickly, the motor was trained by the
environment not to do that again (the repair guy was sent out to change the
behavior of the motor to reduce the probability that it would repeat the
same behavior again (aka fix it)).
So why does the motor correctly raise the door in response to the
prediction of the electric eye today? Because it's past experience taught
it to do it.
The only thing special about "prediction" is that it's a word we like to
use to talk about humans and a word we don't like to use when talking about
machines. And this language usage which is just a social convention makes
the correct understanding of what humans are all that much harder for
people.
> > For it to be able to really
> > "guess" what how some novel situation might unfold, it would need to
> > have some assumptions of the meanings of things it is observing in the
> > situation. Like sounds and lights.
Exactly. The electric eye is observing light levels and it's making a
guess about what the light means. For low levels of light, it makes the
guess that there is no car coming. For a sudden high level of light, it
makes the guess that a car is coming, and sends that guess to the motor as
an electrical signal.
Each new increase in light is an exposure to a novel situation but the
electric eye circuit will always at some point make the prediction that a
car is coming even if the current light is unlike anything it's seen in the
past.
> > Perhaps when it saw just one
> > headlight, and heard no engine noises, it might guess it means there is
> > a bicycle coming and it doesn't need to open the door, etc.
Exactly. A more complex array of sensors can make different guesses based
on different information.
> > Or perhaps it understood the "meaning" of burglars, or more to the
> > point, the meaning of commando masks and pick up "suspicious behaviour"
> > of those who are arriving to mean it shouldn't open itself. And so on.
> > Those all require it to build a worldview of semantical nature.
The electric eye example I gave has a worldview and is using its knowledge
of the world to make a guess about when cars are coming. It's doing the
exact same thing we do when we use our world view to understand how to
react.
Only our world view was created by a combination of nature and nurture. We
have learning hardware in our head as well as reaction hardware, and that
learning hardware caused our reaction hardware to change over time in
response to external stimulus which the critic hardware classified for us
as rewards and punishments.
The garage door by itself doesn't include learning hardware. But if you
include the guys that built the door system and fixed it when it was broken
as one large system, then all that together becomes a learning machine
which is changing over time based on past experience. And this system as a
whole is predicting the future (cars coming) based on past experience (the
electric eye was shown to be a good predictor of cars so it survived for us
to talk about it today). Now this becomes a bit tricky since we are
including humans (which are known to be intelligent learning systems) as
part of the system (like the Chinese room problem). So it's easy to just
push off all the "predicting" on the humans and say what the garage door
isn't doing any predicting, only the human that designed the door predicted
that that design would work well. But you would be wrong to believe it was
that simple.
There are an endless number of examples in the world about how the universe
performed this same trick of "predicting" without the help of humans.
Biological evolution is a prime example. Evolution predicted that having a
body like a human with all it's features would be a good way for the the
body to survive.
Now humans do have this extra power of generating language and the language
can be a prediction about the future. But our tendency to generate such
language can all be explained in terms of past experience showing us that
generating this language was useful to us. And in every case, the
generation of that language happens as a simple reaction, to some current
event. We see dark clouds and we spit out the words "Looks like it's going
to rain". We hear someone say "looks like it's going to rain" and we grab
our raincoat on the way out.
The word "prediction" in the end, is just not a very useful word to help us
understand how humans work, or to help us understand how something like the
garage door system is different from a brain. Or at least, it's not useful
if it gets you confused about how the prediction happens (as a reaction
selected for it's predictive powers by past experience).
Anssi Hyytiainen
>> The fact that this imaginary form manifestates itself from the
>> processes that the neurons are causing doesn't mean the imaginary
>> form IS the neurons, even if you are imagining a neuron itself.
>
> It does actually. That's what happens when you realize there is no
> mind body split and fix all the problems in our language that is
> built on the assumption that there is a mind body split. The brain
> does not represent some imaginary form of the world outside for us as
> much as it _IS_ the world to us. The internal behavior of the brain
> is our world. It's the only world we can ever know.
Have I said something about a mind body split? :o
Everything that is my "conscious mind" is springing from the mechanics
of my body, and mechanics alone, I think I've made my position about
that very clear.
Imagine a neuron. I'm calling it neuron but it could be called "bunch of
atoms" or "piece of brain" or "organic tissue" or whatever. But none of
these expressions, or whatever ideas they bring to your mind, are not
actually "the neuron" as it exists in the real world. By this I don't
mean you are a brain in a vat. I mean, your conscious mind is limited to
mere "ideas" of the world.
The process that brings about your "conscious mind", cannot by mere
"thoughts" it is producing, bring about the "true reality" of neuron per
se. Your "thought" (specific organization of matter/information) only
represent the idea of "a neuron" in some ways. This specific
organization is a mere "expression" which does NOT have the same
"meaning" to anything else but to the process that is your "mind".
And the physical process that is the neuron itself works the opposite
way; The neuron can ONLY "know" of the "true reality" of whatever is
happening to it, in a completely explicit sense. It cannot build any
artificial/conscious "image" of what is happening, because it simply
doesn't hold such organization of matter inside that would bring about a
semantical learning process. It cannot hold such imaginary ideas that my
so-called "mind" is LIMITED to.
So my "mind" is NOT capable of "understanding" the reality of such thing
as "a neuron just fired" when it happens in my brain. I can look at
something that is expressing such an event, or even look at a neuron
itself (which would also just cause an artificial visual image of
reality to my "conscious awareness"), or merely imagine a neuron, but
all of this is just my own semantical view of "understanding a neuron
firing". It is not ACTUALLY a "neuron firing". To hold such an idea
requires that there are neurons firing. None of these firings I am
"aware of" either. I'm just aware of some pattern/expression that they
cause.
So, any process that can be called "conscious mind", cannot possibly
understand in any "true" sense that a neuron is firing.
It is important to get that high in the structure of a self-conscious
mind, instead of just call everything conscious. It is ALWAYS true that
everything is only bunch of quarks "minding their own business", but the
reason I am using such words as consciousness and mind and what have
you, is to describe the sort of mechanism that causes "a mind" to exist
in our brain, in a way that "a mind" doesn't exist in "less
sophisticated" compositions of matter.
> And this is just the problem that type of talk gets you into.
>
> The electric eye example is as much a prediction as what the brain
does but
> it doesn't seem that way because we were taught to talk about human
> behavior using intentional words like "prediction" and "knowledge"
> and "meaning" but we were taught to talk about the operation of a
> machine
using
> extensional words like "reactions".
The electric eye and the predictive process that goes in our brain are
both pure reactions, but the process with the electric eye is not "as
much a prediction as what the brain does". Here, again, prediction is
not something that I consider to be some sort of magical fundamental
thing. It is something that necessarily springs from such a learning
process that I've been describing. "Predictions" of a mind are always
completely semantical, they are not rooted to any "actual knowledge" of
reality.
For the electric eye, the light it is receiving is "actual reality". It
is receiving lights, but it is not having any imaginary/semantical ideas
about what this light means.
For the brain, yes, receiving light is also completely mechanical chain
of reactions, but it is that chain of reactions, that will cause "the
mind" *mechanically to have semantical thoughts* of this experience.
There is so-called "imagination" (not fundamental!). Ideas or images,
artifical expression of reality, whatever you want to call it. These
things occur as patterns. As high level functions.
That's the important difference between when we say a mind "understands
something" or a simpler process "understanding something". The matter of
the fact is, that the mind doesn't truly "understand something" at all!
It just creates an artificial expression of what it is experiencing, and
the simpler process just "understands", without any artificial
expressions being formed.
That was the central point of the original post. Can you accept such a
distinction at all? Keep in mind that ALL distinctions of ANY things are
ultimately semantical.
> So since we describe human behavior using such a different language
> than the language we use to describe machine behavior, it ends up
> leaving us with the obvious conclusion that the two processes are
> substantially different when they aren't.
Of course they aren't substantially different. Nothing is. Everything is
"the same thing" if you want to see it like that.
But we can draw a meaningful difference between "actual reality" and
"model of reality" (mind), can't we?
The model of reality (mind) emerges from actual reality (neurons), but
it is still only expression (thought) of something which exists in
actual reality as an actual entity. The "thought" is just specific
pattern of the "actual neurons" firing.
> Exactly. The electric eye is observing light levels and it's making
> a guess about what the light means. For low levels of light, it
> makes the guess that there is no car coming. For a sudden high level
> of light, it makes the guess that a car is coming, and sends that
> guess to the motor as an electrical signal.
You really don't think there's any value in making a distinction between
something being "implied" to a system, and something being "dictated"?
Even thought something being implied is actually "dictated" at a low
level, the "implying" is a high-level function that occurs only to such
a process that builds imaginary expressions/models of the world, even
though those expressions are also built by mechanical procedure.
The imaginary model is just something that is not rooted to anything
"actual" in the real world at all. That is what makes it imaginary. That
is all we can be conscious of.
So to use such language that the electric eye would be guessing things
is rather odd. It doesn't have such functions that would cause it to
pick up any implied meanings. It just does what it's supposed to. It is
not "clueless" the way our mind is :)
To summarize my view;
- Intention, prediction, consciousness, mind, free-will. None of these
are fundamental.
- Laws of physics are fundamental.
- Laws of physics cause such evolutionary mechanism, that builds such a
composition of matter that is our brain.
- The brain is only a mechanical construction obeying laws of physics
and nothing else.
- The process that the composition of brain and sensory systems cause,
is such that imaginary patterns representing actual reality are being
formed. *Observed reality turns into artificial expression inside the brain*
The expressions are fundamentally "meaningless" patterns representing
actual reality. But because there are such patterns, actual reality is
being artificially modeled, and so predictions about the actual reality
occur.
"I exist" is only such an imaginary "prediction". This imaginary concept
is all that is needed to bring about a conscious experience to such
system. Your "self" is just imaginary!
Rocks are not building artificial expressions of reality. They don't
imagine they exist.
-Anssi
Lucky
>-Anssi
Yep. Right. I am assuming that we need two programs co-operating,
acting
like a single program
(Ever notice how we just assume our mind is "one", and therefore A.I.
will be "one" as well, even though our bodies are made of trillions of
cells, we have two brain hemispheres, and possess both conscious and
unconscious processess?)
One will be responsible for gathering sensory information and sending
such to the second program who is responsible for the structuring it.
This second program will need to have two "bins", one for holding
knowledge not yet processed, and one for holding knowledge already
processed. Hell, we could throw in a third program or more....all that
matters is that they
co-operate so efficiently and be so dependant on each other that they
can no longer be viewed as individual programs, meaning we may have to
specificly program flaws, such as a cascade or domino effect should the
process be interupted or halted. If we think about it, in a sense we
are "turned on" for all of our life, therefore we must also prepare
A.I. for this as well. To turn off A.I. would be the same as killing
it.
This will be incredible freaking hard to do.
-Jason
Anssi Hyytiainen
rejecting your idea, I should mention I just came across some short
writings of Francis Heylighen, who is some sort of professor at "Free
University of Brussels".
He seems to have basically the same idea of what consciousness is (a
high-level process springing from learning and senses and what not), and
he also goes to say that this essentially solves the hard problem;
Pretty clear mechanical description of consciousness:
http://pespmc1.vub.ac.be/CONSCIOU.html
"...many consciousness theorists would claim that such a robot would
still not be conscious, because it would lack what they call "first
person experience" or "qualia". This so-called "hard problem of
consciousness" vanishes if it is considered from a cybernetic point of
view, according to which the property of consciousness is determined by
the robot's organization, not by some mysterious substance, fluid or force."
http://pespmc1.vub.ac.be/HARDPROB.html
http://pespmc1.vub.ac.be/HEYL.html
-Anssi
Lucky
Anssi Hyytiainen wrote:
> btw, since you've been saying that researchers and professors have been
> rejecting your idea, I should mention I just came across some short
> writings of Francis Heylighen, who is some sort of professor at "Free
> University of Brussels".
>
> He seems to have basically the same idea of what consciousness is (a
> high-level process springing from learning and senses and what not), and
> he also goes to say that this essentially solves the hard problem;
>
> Pretty clear mechanical description of consciousness:
> http://pespmc1.vub.ac.be/CONSCIOU.html
Except that my theory is far more simpler! "Francis Heylighen" adds
the usual things that most people claim are necessary for intelligence.
My theory states that only the abilitie to both gather information via
sensory input and comprehend such information is necessary for
intelligence. "Self-Awareness", "Consciousness", etc...are all just
by- products of comprehension.
-Jason
prag
My sex acts are purely mechanical.
Mechanical things can be simulated in computers.
Therefore, I can give up sex
All my sex acts can be simulated on a computer
I will find this just as satisfying as real sex.
Anssi Hyytiainen wrote:
> Chinese Room argument goes as follows:
> ---
> Imagine a native English speaker who knows no Chinese locked in a room
> full of boxes of Chinese symbols together with a book of instructions
> for manipulating the symbols. Imagine that people outside the room send
> in other Chinese symbols which, unknown to the person in the room, are
> questions in Chinese. And imagine that by following the instructions in
> the program the man in the room is able to pass out Chinese symbols
> which are correct answers to the questions. The program enables the
> person in the room to pass the Turing Test for understanding Chinese but
> he does not understand a word of Chinese.
> ---
>
> The point of the above argument is that, like a machine, the "chinese
> room" is following explicit instructions, without understanding the
> *meanings* of the symbols. This argument revolves around the notion that
> we are semantical machines. That we understand MEANINGS in the way that
> an algorithm cannot, since it is just following a mechanical procedure.
>
> I would like to point out, that the true nature of our semantical
> understanding is springing from completely *mechanical* (algorithmic)
> process of learning. I suspect John Searle would appreciate my
> counter-argument:
>
> Semantical understanding is due to mechanical learning:
>
> During our life we continuously learn things. We build "concepts" and
> learn languages. When we learn new concepts, we base them on the
> pre-existing concepts in our "worldview". When we learn new words, we
> either look up their meaning from a dictionary through other words, or
> understand the meaning of the new word through pre-learned concepts.
>
> The matter of the fact is that in our conscious mind:
> Words only have meaning relative to other words.
> Concepts only have meaning relative to other concepts.
>
> And this is so, because our learned worldview is not rooted to any
> explicit knowledge of the world. It is at heart, a self-supporting
> circle of beliefs:
>
> When you were born, you obviously didn't know anything about anything.
> You didn't know any words, you didn't understand any concepts. After
> being brought into the world, your brain, a self-organizing learning
> system, started a rigorous but mechanical process of learning, It tries
> to seek some sense to all the strange sensory input.
>
> Whatever assumptions it placed into the worldview that seemed to give
> meaning to other assumptions, and vice versa, will become the
> self-supporting root of your worldview. Don't underestimate the learning
> ability of a fresh brain. It's a friggin' sponge I tell you! In fact,
> that "you exist" is also just a semantical assumption you made at some
> point of your life. It was only after that assumption that you could
> store any memories that you interpetated as something occuring to you.
> (That is why there is infant amnesia, and that is why you didn't really
> even exist until your brain made assumptions as to what is "existence",
> and that "I probably also do exist")
>
> So while our learning process is completely mechanical, we are the
> product of our experiences, and we don't truly understand anything in an
> explicit sense. We are forced to see everything through semantics. Even
> math is semantical to us. 6 x 7 = 42 just because we agreed on the rules
> of how it is so. Counting such a thing in our head is not algorithmic
> process like it is for a computer (if it was, you would not calculate
> anything consciously).
>
> We build a world-view this way, because it allows us to understand how
> things work. This is same as understanding the meaning of things. When
> we understand how things work, we can predict our environment, and
> survive with our wits.
>
> Because of this, we are forced to *interpetate the meaning* of
> EVERYTHING we see around us.
>
> Human communication can never be understood; it must always be interpetated.
> Computer communication can never be interpetated; it must always be
> interpetated.
>
> But since this learning process IS explicit, then if we build a machine
> that learns the same way, through explicit algorithm, it will not
> consciously understand this learning process either, but IT WILL
> UNDERSTAND SEMANTICS.
>
> The problem with the chinese room argument is simply that it assumes no
> learning takes place. The instruction table must not be given into the
> room. The room must have ability to learn, and it must have "sensory
> systems" that will give it information about the world that it is
> supposed to try to make sense of. Then it will become to understand chinese.
>
> It must be noted that while the worldview is a product of our
> experiences, its interpetation is also a mechanical process.
> Intentionality then, is a kind of illusion. Although you interpetate the
> world in a way that you are consciously making decisions, and have free
> will, the matter of the fact is that *you can only think of what you can
> think of*, and your actions are in fact mechanically restricted by your
> worldview, at all moments. You cannot base your decisions on anything
> you haven't yet learned.
>
> Anyone aspiring to build a truly conscious AI, should not put is
> attention into how to program behaviour directly, but into how to make
> the system learn in an open-ended fashion, without any true
> understanding of anything.
>
> If anyone knows Searle's e-mail address, I would be very interested to
> hear what he might think about the above.
>
> -Anssi
Anssi Hyytiainen
Hmmmm, as far as I can tell he is not adding any multiple things
together, he is just describing how these things come to exist in a
learning machine.
-Anssi
Anssi Hyytiainen
-Anssi
prag
I'm talking no wet-wear, none.
When you figure out what I am taking about I know your are going to
start claiming that you can still engage old in-out after you are
uploaded to a computer, but Searle would argue that you can't be
uploaded to a *simulation*. Searle would grant that uploading is
possible in principle and he would grant that hi-fidelity simulations
of the sex act are possible on a computer, but he would claim that you
cannot be uploaded to a simulation.
Anssi Hyytiainen
That's really the only reason why I was hoping Searle might be wrong.
But maybe Searle is wrong after all. He doesn't know what we might be
missing. In fact, let us quiet down and pray that Searle is wrong, just
in case.
"Our father, who art in heaven. If there's one thing we want from you,
please let Searle be mistaken, and bless us with hi-fi porn of great
proportions. We'll probably want it every day. Let it be just like real
sex, only with hotter chicks. Thanks for listening. Ps. the sooner, the
better"
-Anssi
Curt Welch
> No, No, I don't mean THAT!
>
> I'm talking no wet-wear, none.
>
> When you figure out what I am taking about I know your are going to
> start claiming that you can still engage old in-out after you are
> uploaded to a computer, but Searle would argue that you can't be
> uploaded to a *simulation*. Searle would grant that uploading is
> possible in principle and he would grant that hi-fidelity simulations
> of the sex act are possible on a computer, but he would claim that you
> cannot be uploaded to a simulation.
Well, I don't see what this has to do with what Anssi was writing. But on
the issue of "uploading"...
If you believe that we are nothing but machines, and that there is no
mind-body split, the body is all there is to us. The ides of uploading a
body is no different than trying to upload a computer.
In other words, you can't do it. All you can do, is build a clone that
works the same way. Computers are designed to be very easily cloned. You
can just copy the software on a machine and clone the function even if the
computer body is physically very different. But if the software was
controlling a robot body, then you would also need to clone the body to a
very high degree in order to duplicate the function of the software just
like you would have to duplicate the function of all the I/O devices to
duplicate the function of computer software.
Human brains however are not digital. They use a ton of analog components
such as synapse weights and the length of nerves which change the timing of
signals which changes the function of the brain. There are probably also
effects created by the speed at which chemicals flow though the brain which
is a complex function of the physical layout of the parts and things like
the path of blood veins. So to duplicate a human brain (and end up with
the same personality/person after the duplication), you would have to clone
the function of all these analog components to a very high degree of
accuracy. It's not something you could easily or practically clone in
digital software.
And of course, the brain is there to control the body it's attached to, and
has spend a life time learning how it's body works, so you would need to
duplicate not just the brain, but the entire body in order to end up with
something that was in effect, the same basic person.
You could attempt to create a computer simulation of a human body, but the
amount of computation required to simulate the whole human body to the
level needed to make it work correctly (to duplicate the mind of the
original body) would likely require so much computing power as to make it
totally impractical (you might for example need a computer the size of the
earth to do it).
And when you were done, the person would not be uploaded to the computer,
all you would have done is built a clone. If you didn't want the original,
you would just have to kill it.
When we learn to duplicate the function of the brain in a computer however,
we won't need to do all that simulation. Just like when we build a clone
of a computer, the clone machine doesn't need to simulate the complex
analog function of every transistor in the first computer. It only needs
to duplicate the digital functions of the first. So with these machines,
like with normal computers, you will be able to clone the function, just by
copying the software. But once again, it's not as if the "mind" exists as
something separate from the body and could be transferred to a new body.
All we can ever do is build clones.