4 Tests Reveal Bing (GPT 4) ≈ 114 IQ (last test is nuts)

[John Clark]

Forget the Ukraine war, forget climate change, forget Donald Trump, I now think GPT-4 is by far the most world shaking event and the most underreported one.  Many of us have been talking about the singularity for decades, but now it looks like we're on its doorstep. You've got to look at this video!

4 Tests Reveal Bing (GPT 4) ≈ 114 IQ (last test is nuts)

John K Clark    See what's on my new list at  Extropolis

cdk

[spudb...@aol.com]

I have heard this prediction over that last week from other experts, than JC. No to 2050/2045 and no to 2030. 

This also came out in the last 48 hours. Kurzweil is an inventor, not a scientist. Having said that he was an inventor.

This might be of interest to JC? Or not?

Humans will achieve immortality in 7 years, says futurist

Humans will achieve immortality in 7 years, says futurist (interestingengineering.com)

My wandering thought is, IF TRUE (eyes rolling, tongue in cheek), America executes prisoners for capital murder. So, legally, if a murderer died 50 years ago for a capital crime, does that mean, once, revived, they are no longer liable because of Double Jeopardy and that their victims will also be revived? A Civil Case then??

The other thought was a 2nd Hitler War, with Joe Stalin and Mao teamed up against the rest of us? Churchill, George Clemenceau, William Tecumseh Sherman (strategy that beat The South), José Norton de Matos, and a now militarized Ganndhi????? If we're all electronic/quantum returnees, how would these wars be fought? 

--
You received this message because you are subscribed to the Google Groups "Everything List" group.
To unsubscribe from this group and stop receiving emails from it, send an email to everything-li...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/everything-list/CAJPayv0SOiZQ3fLU-8fQDX5Zfs6%3Dm5koRbJzRV%2Bdg6E-a-mm%3DA%40mail.gmail.com.

[John Clark]

On Thu, Mar 16, 2023 at 4:50 PM <spudb...@aol.com> wrote:

> America executes prisoners for capital murder. So, legally, if a murderer died 50 years ago for a capital crime, does that mean, once, revived, 

To my knowledge no executed prisoner has ever been cryogenically preserved, however Joseph Paul Jernigan was executed by lethal injection in 1981 and he became part of the "Visible Human Project". His body was sliced into 1871 1 millimeter thick slices. and each slice was then photographed with a very high resolution camera. I've wondered if there was enough information preserved in those photographs to upload him, probably not but maybe. You can watch a one minute video of a journey through Mr. Jernigan's body here.

The visible human project - Male (HD)

A few years later they took even higher resolution photographs of a woman who died of a heart attack and they used even thinner slices, only 0.33 millimeters thick

Visual Human Project Female

> they are no longer liable because of Double Jeopardy and that their victims will also be revived? A Civil Case then??

As I've said, I'm not a lawyer.  

John K Clark    See what's on my new list at  Extropolis

ws8

u6c

[spudb...@aol.com]

Yes. I actually had, long ago, somebody who contributed their mortal remains to science, only, to have one delicious sub-slice winding up in our computer graphics class. A diameter slice of their skull. The computer graphics class was wasted on myself since artwork was better left for the artists. For the contributor, it was no great contribution to science since scientists can also be untrustworthy. 

I have a name for the slice I encountered long ago. I thought this being could also be a great sidekick, if I could write the comic lines for the voice? 

[spudb...@aol.com]

To get to the point, I did advocate for a bit of skepticism for claiming consciousness for a computer system, and the retort was from JC that essentially, we cannot even define what makes a human conscious, and I am going with an au contraries', Pierre! I took me under 10 min to locate a worthy article submitted for JC's criticisms. 

Here tis'

What Neuroscientists Think, and Don't Think, About Consciousness - PubMed (nih.gov)

So, we are much closer to understand human consciousness. I am ask to to put the same effort into how a network developed this in so little time. Our our analog chips so mighty in 2022-3???

-----Original Message-----
From: John Clark <johnk...@gmail.com>

To: spudb...@aol.com
Cc: everyth...@googlegroups.com <everyth...@googlegroups.com>
Sent: Thu, Mar 16, 2023 5:55 pm

--
You received this message because you are subscribed to the Google Groups "Everything List" group.
To unsubscribe from this group and stop receiving emails from it, send an email to everything-li...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/everything-list/CAJPayv2neEP50SN7NCGRqtm1uzwNaE668DFu4-QdQa-OGFei8A%40mail.gmail.com.

[Telmo Menezes]

Am Do, 16. Mär 2023, um 18:30, schrieb John Clark:

Forget the Ukraine war, forget climate change, forget Donald Trump, I now think GPT-4 is by far the most world shaking event and the most underreported one.  Many of us have been talking about the singularity for decades, but now it looks like we're on its doorstep. You've got to look at this video!

4 Tests Reveal Bing (GPT 4) ≈ 114 IQ (last test is nuts)

One crucial question here is: did the GPT 4 training set include tests like the ones being solved?

Telmo

John K Clark    See what's on my new list at  Extropolis

cdk

--

You received this message because you are subscribed to the Google Groups "Everything List" group.

To unsubscribe from this group and stop receiving emails from it, send an email to everything-li...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/everything-list/CAJPayv0SOiZQ3fLU-8fQDX5Zfs6%3Dm5koRbJzRV%2Bdg6E-a-mm%3DA%40mail.gmail.com.

[Stathis Papaioannou]

On Fri, 17 Mar 2023 at 07:37, spudboy100 via Everything List <everyth...@googlegroups.com> wrote:

To get to the point, I did advocate for a bit of skepticism for claiming consciousness for a computer system, and the retort was from JC that essentially, we cannot even define what makes a human conscious, and I am going with an au contraries', Pierre! I took me under 10 min to locate a worthy article submitted for JC's criticisms. 

Here tis'

What Neuroscientists Think, and Don't Think, About Consciousness - PubMed (nih.gov)

So, we are much closer to understand human consciousness. I am ask to to put the same effort into how a network developed this in so little time. Our our analog chips so mighty in 2022-3???

What produces consciousness is consciousness-associated behaviour. If it’s EM fields (it probably isn’t) then EM fields would have some effect on behaviour, and if this effect could be replicated by some other means, the consciousness would also be replicated.

--

Stathis Papaioannou

[John Clark]

On Fri, Mar 17, 2023 at 3:45 AM Telmo Menezes <te...@telmomenezes.net> wrote:

>> Forget the Ukraine war, forget climate change, forget Donald Trump, I now think GPT-4 is by far the most world shaking event and the most underreported one.  Many of us have been talking about the singularity for decades, but now it looks like we're on its doorstep. You've got to look at this video!
4 Tests Reveal Bing (GPT 4) ≈ 114 IQ (last test is nuts)

> One crucial question here is: did the GPT 4 training set include tests like the ones being solved?

I don't know, GPT-4 is trained on a huge amount of data so probably, but why is that a crucial question?  When human beings take an IQ test that is almost certainly NOT the first test they've ever had, and like GPT-4 humans are also trained on a huge amount of data, without it neither you nor GPT-4 would even know how to read the questions.  And speaking of that, in one of those 4 tests humans had deliberately written the questions in such a convoluted way that it was difficult to even know what the question was, much less find the answer,  but GPT-4 got it right nevertheless.  And although GPT-4 can input graphs and diagrams the version that was taking the 4 different IQ tests could not, so whenever an IQ problem contained one of those GPT-4 was automatically marked as getting the answer wrong, and yet even with that severe handicap and even with being unable to contact the Internet it STILL managed to get an IQ of 114 !  And if you had told me in 2021 that a machine would be capable of doing that in 2023 I would've said you were crazy.  

 John K Clark    See what's on my new list at  Extropolis

8gw

9x

[Telmo Menezes]

Am Fr, 17. Mär 2023, um 12:52, schrieb John Clark:

On Fri, Mar 17, 2023 at 3:45 AM Telmo Menezes <te...@telmomenezes.net> wrote:

>> Forget the Ukraine war, forget climate change, forget Donald Trump, I now think GPT-4 is by far the most world shaking event and the most underreported one.  Many of us have been talking about the singularity for decades, but now it looks like we're on its doorstep. You've got to look at this video!
4 Tests Reveal Bing (GPT 4) ≈ 114 IQ (last test is nuts)

> One crucial question here is: did the GPT 4 training set include tests like the ones being solved?

I don't know, GPT-4 is trained on a huge amount of data so probably, but why is that a crucial question? 

Well, this is Machine Learning 101. If you train a model, it will always perform better in corpus. Often MUCH better. That is why it is Machine Learning 101 to divide your dataset into training and testing (usually a 66%/33% or 75%/25% split). Try to publish a ML learning paper where you evaluate your model based on in corpus data and see what they tell you... I am simply insisting on the conventional scientific standard of the field to evaluate this "≈ 114 IQ" claim.

When human beings take an IQ test that is almost certainly NOT the first test they've ever had, and like GPT-4 humans are also trained on a huge amount of data, without it neither you nor GPT-4 would even know how to read the questions.

Yes, but GPT-4 and human brains are very different things. GPT-4 has superhuman memory capabilities and almost certainly subhuman reasoning capabilities. So saying that "ah yes, but humans also see IQ tests before taking one" is comparing apples to oranges, firstly beacuse GPT-4 relies much more on brute force memorization and secondly because its training corpus can be incredibly exhaustive. I strongly suspect that GPT-4 is trained with essentially all publically available text ever produced. I bet that it would take many human lifetimes to actually read the entire training datatset of GPT-4.

I have a leak of the training data of GPT-2 and that already appears to be the case.

  And speaking of that, in one of those 4 tests humans had deliberately written the questions in such a convoluted way that it was difficult to even know what the question was, much less find the answer,  but GPT-4 got it right nevertheless.  And although GPT-4 can input graphs and diagrams the version that was taking the 4 different IQ tests could not, so whenever an IQ problem contained one of those GPT-4 was automatically marked as getting the answer wrong, and yet even with that severe handicap and even with being unable to contact the Internet it STILL managed to get an IQ of 114 !  And if you had told me in 2021 that a machine would be capable of doing that in 2023 I would've said you were crazy.

Again, it is important to understand what exactly GPT-4 is doing. It is certainly impressive, but it is not the same thing as a human being taking an IQ test, and this is important because it has profound implications on what an approach like GPT-4 can be generalized to do and what its fundamental limitations are.

Telmo

 

 John K Clark    See what's on my new list at  Extropolis

8gw

9x

--

You received this message because you are subscribed to the Google Groups "Everything List" group.

To unsubscribe from this group and stop receiving emails from it, send an email to everything-li...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/everything-list/CAJPayv2-GRdQ5s%2BQ7DbVnR54m1ecwV4ukQMs0_W1%2BnHaH2k0Qg%40mail.gmail.com.

[John Clark]

On Fri, Mar 17, 2023 at 8:26 AM Telmo Menezes <te...@telmomenezes.net> wrote:

> Well, this is Machine Learning 101. If you train a model, it will always perform better

Well yes, if a machine couldn't learn then it wouldn't be intelligent!

>> When human beings take an IQ test that is almost certainly NOT the first test they've ever had, and like GPT-4 humans are also trained on a huge amount of data, without it neither you nor GPT-4 would even know how to read the questions.

> Yes, but GPT-4 and human brains are very different things.

They are similar in some ways, they both process information, and they are different in other ways, one processes information using carbon chemistry and the other processes information using silicon electronics.

> GPT-4 has superhuman memory capabilities

Yes

 

> and almost certainly subhuman reasoning capabilities.

If so I've seen no evidence of it, I have however seen evidence that the opposite is true.  

> I bet that it would take many human lifetimes to actually read the entire training datatset of GPT-4.

I'm sure that's true. Am I supposed to think less of  GPT-4 because of that?

Again, it is important to understand what exactly GPT-4 is doing. It is certainly impressive, but it is not the same thing as a human being taking an IQ test,

It's the same thing if you treat both humans and machines as black boxes and concentrate on what they do. Like it or not that's the only way we can deal  with our fellow human beings that we encounter in everyday life, we have no way of knowing what's going on inside their head, all we can do is observe their behavior. Maybe Einstein was an idiot but he just had an ability to push a pen in such a way that he produced brilliant physics papers, but nobody believes that; instead we would say if somebody could write physics papers that were as brilliant as Einstein's then that person would be as smart as Einstein.  

It's interesting that until a few years ago the Turing Test was not very controversial because most thought it would be centuries before a machine could pass it, and many proclaimed a machine would never be able to pass it, but now that a machine has indeed passed it they say the Turing Test is not important, even though they personally still use the Turing Test a 1000 times a day whenever they judge the conscious state of one of their fellow human beings. Actually if GPT-4 really wanted to fool somebody into thinking it was a human being it would have to dumb itself down.  

John K Clark    See what's on my new list at  Extropolis

h5z

8bm

[Jason Resch]

On Thu, Mar 16, 2023, 6:37 PM spudboy100 via Everything List <everyth...@googlegroups.com> wrote:

To get to the point, I did advocate for a bit of skepticism for claiming consciousness for a computer system, and the retort was from JC that essentially, we cannot even define what makes a human conscious, and I am going with an au contraries', Pierre! I took me under 10 min to locate a worthy article submitted for JC's criticisms. 

Here tis'

What Neuroscientists Think, and Don't Think, About Consciousness - PubMed (nih.gov)

So, we are much closer to understand human consciousness. I am ask to to put the same effort into how a network developed this in so little time. Our our analog chips so mighty in 2022-3???

Neurologists know that neurons and neural activity is correlated with consciousness, but for the most part their understanding stops there, (and by their own admission.)

I would say neurologists are almost in the worst position to understand consciousness as they look at it from the lowest level, the neurons. This is like trying to decipher a word processor program by looking at the patterns of electrical impulses in the circuits of a computer's CPU.

Here are some quotes about our complete lack of understanding of consciousness and the disappointment regarding what help neurology has offered (emphasis mine):

“How it is that anything so remarkable as a state of consciousness comes about as a result of irritating nervous tissue, is just as unaccountable as the appearance of Djin when Aladdin rubbed his lamp.”

-- Thomas Huxley in " “Lessons in Elementary Psychology,” (1866)

“An electron is neither red nor blue nor any other colour; the same holds for the proton, the nucleus of the hydrogen atom. But the union of the two in the atom of hydrogen, according to the physicist, produces electromagnetic radiation of a certain discrete array of wavelengths. The homogenous constituents of this radiation, when separated by a prism or an optical grating, stimulate in an observer the sensations of red, green, blue, violet by the intermediary of certain physiological processes, whose general character is sufficiently well known to assert that they are not red or green or blue, in fact that the nervous elements in question display no colour in virtue of their being stimulated; the white or gray the nerve cells exhibit whether stimulated or not is certainly insignificant in respect of the colour sensation which, in the individual whose nerves they are, accompanies their excitation.”

-- Erwin Schrödinger in "Mind and Matter" (1958)

“Few questions have endured longer or traversed a more perplexing history than this, the problem of consciousness and its place in nature. Despite centuries of pondering and experiment, of trying to get together two supposed entities called mind and matter in one age, subject and object in another, or soul and body in still others, despite endless discoursing on the streams, states, or contents of consciousness, of distinguishing terms like intuitions, sense data, the given, raw feels, the sensa, presentations and representations, the sensations, images, and affections of structuralist introspections, the evidential data of the scientific positivist, phenomenological fields, the apparitions of Hobbes, the phenomena of Kant, the appearances of the idealist, the elements of Mach, the phanera of Peirce, or the category errors of Ryle, in spite of all of these, the problem of consciousness is still with us.”

-- Julian Jaynes in "The Origin of Consciousness in the Breakdown of the Bicameral Mind" (1976)

“We know that brains are the de facto causal basis of consciousness, but we have, it seems, no understanding whatever of how this can be so. It strikes us as miraculous, eerie, even faintly comic. Somehow, we feel, the water of the physical brain is turned into the wine of consciousness, but we draw a total blank on the nature of this conversion. Neural transmissions just seem like the wrong kind of materials with which to bring consciousness into the world, but it appears that in some way they perform this mysterious feat. The mind-body problem is the problem of understanding how the miracle is wrought, thus removing the sense of deep mystery. We want to take the magic out of the link between consciousness and the brain.”

-- McGinn “Can we solve the mind body problem?” (1989)

“IT IS REMARKABLE that most of the work in both cognitive science and the neurosciences makes no reference to consciousness (or 'awareness'), especially as many would regard consciousness a the major puzzle confronting the neural view of the mind and indeed at the present time it appears deeply mysterious to many people.”

-- Francis Crick in "Towards a neurobiological theory of consciousness" (1990)

“At the time, I uncritically accepted the view that the troublesome phenomenal, or “what it is like,” aspect of experiences had nothing to do with their representational contents, and I supposed that neurophysiology would ultimately tell the full story. In the course of reflecting on this pair of assumptions in later years, I Came to think that I had made a serious mistake. Not only are the phenomenal or felt aspects of our mental lives representational but also (relatedly) they are not even in the head at all. So, neurophysiology certainly will not reveal to us what it is like to smell or skunk or to taste a fig. Look at the neurons for as long as you like, and you will not find phenomenal consciousness.”

-- Michael Tye in "Ten Problems of consciousness" (1995)

“Consciousness poses the most baffling problems in the science of the mind. There is nothing that we know more intimately than conscious experience, but there is nothing that is harder to explain. All sorts of mental phenomena have yielded to scientific investigation in recent years, but consciousness has stubbornly resisted. Many have tried to explain it, but the explanations always seem to fall short of the target. Some have been led to suppose that the problem is intractable, and that no good explanation can be given.”

-- David Chalmers in "Facing Up to the Hard Problem" (1996)

“We should therefore not expect the search for a neural correlate of consciousness to lead to the holy grail of a universal theory. We might expect it to be valuable in helping us to understand consciousness in specific cases, such as the human case: learning more about the processes underlying awareness will certainly help us understand the structure and dynamics of consciousness, for example. But in holding up the bridge from physical processes to conscious experience, preexperimental coherence principles will always play a central role”

-- David Chalmers in "The Conscious Mind" (1996)

“Two decades later, we know an astonishing amount about the brain: you can’t follow the news for a week without encountering at least one more tale about scientists discovering the brain region associated with gambling, or laziness, or love at first sight, or regret – and that’s only the research that makes the headlines. Meanwhile, the field of artificial intelligence – which focuses on recreating the abilities of the human brain, rather than on what it feels like to be one – has advanced stupendously. But like an obnoxious relative who invites himself to stay for a week and then won’t leave, the Hard Problem remains. When I stubbed my toe on the leg of the dining table this morning, as any student of the brain could tell you, nerve fibres called “C-fibres” shot a message to my spinal cord, sending neurotransmitters to the part of my brain called the thalamus, which activated (among other things) my limbic system. Fine. But how come all that was accompanied by an agonising flash of pain? And what is pain, anyway?” 

-- Oliver Burkeman in “Why can’t the world’s greatest minds solve the mystery of consciousness?” (2015)

I think any satisfying answer must involve all the levels of processing that sit between the top level user-interface of consciousness (which we see and which presents the mystery), and the lowest level of the neural circuitry, which we also see. But between these two layers is a vast bridge of various levels of processing, processing of neurons and dendrites, processing of neocortical columns, algorithms in the cerebellum, processing of sub-brain regions in the visual cortex for recognizing colors, shapes, patterns, motion, faces, etc., the whole regions such as the complete visual cortex, whole brain hemispheres, the complete brain of both hemispheres and connection via the corpus callosum. We might make the analogy between the brain and a cell phone, where again we have the UI presented to us on the screen, and the circuitry at the bottom. In the middle layers are the machine code, the system calls, operating system kernel, functions, routines, modules, sub processes, processes, applications, etc. which finally work their way up to presenting a screen with buttons, text and images. All the meat required for understanding exists in the middle layers. Analyzing the top-most and/or the bottom-most layers, while ignoring the middle is sure to lead to bafflement.

As neurology works its way up to building a complete map of neural regions and functioning, we will know how the brain does what it does, but will that explain consciousness? Here philosophers disagree. Some like Dennett say that is as far as we can go and that will answer all the questions we have about consciousness. Others like Chalmers say that will still leave the "hard problem" unresolved.

I see merit in both aspects of their argument. I agree with Dennett that consciousness is nothing other than awareness. But I also agree with Chalmers that even with such an objective and complete brain map, there will remain some things that are unexplainable/shareable (in my opinion due in part to similar reasons as Gödelian incompleteness). First-person experiences are not explainable in third-person terms and can only be understood/experienced/known by being the system that has that particular experience.

Jason

To view this discussion on the web visit https://groups.google.com/d/msgid/everything-list/382281143.709221.1679006249871%40mail.yahoo.com.

[John Clark]

On Thu, Mar 16, 2023 at 6:37 PM <spudb...@aol.com> wrote:

> I took me under 10 min to locate a worthy article submitted for JC's criticisms. 

I'll bet it took you less than 10 seconds, you popped two or three buzzwords into Google  and then you picked the first one that came up and sent it to the list without reading a word of it. You didn't miss much, it didn't increase my understanding of consciousness one bit, but you claim the article's insights brought us "much closer to understanding human consciousness" , well that part must've been written in invisible ink because I sure didn't see it.

> Here tis [drum roll]

> What Neuroscientists Think, and Don't Think, About Consciousness - PubMed (nih.gov)

 Unlike you I did  read the article, or at least I read the abstract, and based on that I flat out refuse  to waste my time by reading the entire worthless thing, it is after all the purpose of abstracts.

> The approach the majority of neuroscientists take to the question of how consciousness is generated, it is probably fair to say, is to ignore it

Yes it is fair to say that a majority of neuroscientists ignore consciousness and they do so because it's the rational thing to do, it would be silly to divert finite mental resources from fantastically productive intelligence research, especially now when so many dramatic discoveries are being made in that area,  to a moribund field like consciousness research that has not advanced one nanometer in the last thousand years and will not do any better in the next thousand.  

> Neuroscience has furnished evidence that neurons are fundamental to consciousness;

And how did neuroscientists figure that out? By observing that when neurons behave in a certain way organisms behave in a certain way and they guess, I repeat they guess, that when organisms behave in that certain way then their consciousness must be  in a certain state. I personally think that is a good guess, it's the same guess we make when we  determine that  our fellow human beings are not conscious when they are sleeping or under anesthesia or dead.

And then the author starts babbling about  

> staggeringly complex system of electromagnetic field"

A keen grasp of the obvious. Everything except neutrinos and Dark matter Interacts with the electromagnetic field.  

> The EM field literally manifests the computations, or signaling, or information processing/activities performed 

Both computers and neurons produce electromagnetic fields. So what? What does that have to do with consciousness? The author never says, he just maintains that somehow electromagnetic fields produce consciousness and of course he provides no evidence to support his belief.   

John K Clark    See what's on my new list at  Extropolis

txew
w

[spudb...@aol.com]

EM fields might indeed have an effect on the human  neuro-system. I am merely asking how would consciousness develop, unplanned in a server farm? 

Humans, yes. NIMH » Brain Stimulation Therapies (nih.gov)

In a computer? Maybe yes.

A theory of consciousness from a theoretical computer science perspective: Insights from the Conscious Turing Machine | PNAS

If this guy is correct, then simply learning grants consciousness.   

Learning May Be the Key to the Evolution of Consciousness | Psychology Today

I have questions, but because I do, doesn't make me the arbiter of computer consciousness. 

I just don't choose to blow past it so readily as JC does. 

-----Original Message-----
From: Stathis Papaioannou <stat...@gmail.com>
To: everyth...@googlegroups.com
Sent: Fri, Mar 17, 2023 4:23 am
Subject: Re: 4 Tests Reveal Bing (GPT 4) ≈ 114 IQ (last test is nuts)

--
You received this message because you are subscribed to the Google Groups "Everything List" group.
To unsubscribe from this group and stop receiving emails from it, send an email to everything-li...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/everything-list/CAH%3D2ypXKDU7uDj3UHaZXO-0pExsLgbANTcdM76KctK51PvCLSw%40mail.gmail.com.

[spudb...@aol.com]

The authors are professionals. You hold that your knowledge base is is greater than the authors?  EM surely could be the root of how humans get conscious, and that could easily be the quantum field effect, no, not quantum computing, but QFT. Like carbon/water QFET's. 

Do quantum effects play a role in consciousness? – Physics World

We still both for animals and machines need a working analysis. The How questions. 

You're not curious about how animals and machinery (if you are correct?) attained consciousness? 

-----Original Message-----
From: John Clark <johnk...@gmail.com>
To: spudb...@aol.com
Cc: everyth...@googlegroups.com <everyth...@googlegroups.com>
Sent: Fri, Mar 17, 2023 1:34 pm
Subject: Re: 4 Tests Reveal Bing (GPT 4) ≈ 114 IQ (last test is nuts)

[spudb...@aol.com]

I am far, less, the philosopher then you are. All this peasant (me!) requires for both animals and machines is a basic mechanical, cause + effect diagram on how both sets attained self-awareness? Call it a working theory. 

It's remarkable that we attained consciousness and even more remarkable that a server farm could do so. 

If one is a pantheist, then I suppose one sees consciousness in everything, being, "as right as rain." I have no objection to that view either, because maybe the pantheists are correct or will be found so? 

Now, by choice, would I prefer to have a Turning-surpassable computer, or something that unconsciously, churns out wonderful technologies for humanity? I'll pick the later, because we have 8 billion people to chat with on this world, and I choose to chat with people. I personally, would like to chat with my fellow humans about the new, asteroid mining craft GPT6 just produced, but so far, there's zero in the news about that. 

This, comes from my values, but it's non-obligatory that all humans need value this as a first. Each to their own. 

To view this discussion on the web visit https://groups.google.com/d/msgid/everything-list/CA%2BBCJUju2F-JRy%2B0Updaqw879kbQAfj8WBkQ7z36_%3DLFFnAZ0g%40mail.gmail.com.

[smitra]

I'll be convinced if they succeed making such a system do original
research in, say, theoretical physics or mathematics and get lots of
high quality research results published that gets los of citations for
groundbreaking work, rather than small improvements of details of
existing work. If the next Feynman, Einstein, or John on Neumann is an
AI system then skeptics can continue to argue that the system is still
just a dumb computer, but the world will then have moved on with those
AI systems becoming an ever more important part of society.

I'm not convinced that the present GPT systems are demonstrating
human-level intellect. Our brains are very powerful computers that we
can use in a very inefficient way to do academic work with. If the brain
of a lizard were fully dedicated to analyze and process texts instead of
controlling its body to react optimally to its environment, it would
likely also perform extremely well compared to these GPT systems.

The way one would be able to see that the system despite performing
extremely well does not have the intellectual capabilities of a human
being, would be to follow up on gaps in its knowledge and see if it can
learn from its mistakes and master new subjects.

So, in the video we see that it got a question wrong because it thought
that 33 is a prime number. I would be more impressed by a system that
may make many more mistakes like that than this GPT system made, but
where there is a follow-up conversation where the mistakes are pointed
out and the system shows that it has learned and then gets similar
questions that it would previously have gotten wrong given the previous
answers, correct.

Saibal

On 16-03-2023 18:30, John Clark wrote:
> Forget the Ukraine war, forget climate change, forget Donald Trump, I
> now think GPT-4 is by far the most world shaking event and the most
> underreported one. Many of us have been talking about the singularity
> for decades, but now it looks like we're on its doorstep. You've got
> to look at this video!
>

> 4 Tests Reveal Bing (GPT 4) ≈ 114 IQ (last test is nuts) [1]
>
> John K Clark See what's on my new list at Extropolis [2]
>
> cdk

>
> --
> You received this message because you are subscribed to the Google
> Groups "Everything List" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to everything-li...@googlegroups.com.
> To view this discussion on the web visit

> https://groups.google.com/d/msgid/everything-list/CAJPayv0SOiZQ3fLU-8fQDX5Zfs6%3Dm5koRbJzRV%2Bdg6E-a-mm%3DA%40mail.gmail.com
> [3].
>
>
> Links:
> ------
> [1]
> https://www.youtube.com/watch?v=xFvDJnf0GXs&amp;list=PLYXp_rV1HrBAOZqPJTOSo91275hKQrfpl&amp;index=13
> [2] https://groups.google.com/g/extropolis
> [3]
> https://groups.google.com/d/msgid/everything-list/CAJPayv0SOiZQ3fLU-8fQDX5Zfs6%3Dm5koRbJzRV%2Bdg6E-a-mm%3DA%40mail.gmail.com?utm_medium=email&utm_source=footer

[Telmo Menezes]

Am Fr, 17. Mär 2023, um 14:11, schrieb John Clark:

On Fri, Mar 17, 2023 at 8:26 AM Telmo Menezes <te...@telmomenezes.net> wrote:

> Well, this is Machine Learning 101. If you train a model, it will always perform better

Well yes, if a machine couldn't learn then it wouldn't be intelligent!

Ok John, I have an algorithm here that, can 100x times your money every month in the derivatives market. I can demonstrate this by running it on the training data. I will sell it to you for 100K. An incredible bargain, considering that you can become the richest man in the planet in just a few months. Interested?

>> When human beings take an IQ test that is almost certainly NOT the first test they've ever had, and like GPT-4 humans are also trained on a huge amount of data, without it neither you nor GPT-4 would even know how to read the questions.

> Yes, but GPT-4 and human brains are very different things.

They are similar in some ways, they both process information, and they are different in other ways, one processes information using carbon chemistry and the other processes information using silicon electronics.

Other ways in which they are different are also that: one is a Turing-complete self-organizing general learning algorithm with a highly evolved utility function, and the other is a model trained with gradient descent and a huge amount of data to predict the likelihood of words given a context.

Huge progresses is being made, but we are not at the human level of generality of intelligence and autonomy. Not even close. I am sure we can get there, but I would argue that too much hype too early hurts the cause. That's how we got the first AI winter.

> GPT-4 has superhuman memory capabilities

Yes

 

> and almost certainly subhuman reasoning capabilities.

If so I've seen no evidence of it, I have however seen evidence that the opposite is true.  

I fear that you are falling for the very human bias (I fall for it so many times myself) of seeing what you want to see.

> I bet that it would take many human lifetimes to actually read the entire training datatset of GPT-4.

I'm sure that's true. Am I supposed to think less of  GPT-4 because of that?

No, but you are supposed to remain a scientist and keep applying the same fundamental machine learning principle that brought us all the way to GPT-4:

A machine learning system can only be objectively evaluated by applying it to data that was not used to train it. This is the only way to distinguish between true generalization (learning) and over-fitting.

Again, it is important to understand what exactly GPT-4 is doing. It is certainly impressive, but it is not the same thing as a human being taking an IQ test,

It's the same thing if you treat both humans and machines as black boxes and concentrate on what they do. Like it or not that's the only way we can deal  with our fellow human beings that we encounter in everyday life, we have no way of knowing what's going on inside their head, all we can do is observe their behavior. Maybe Einstein was an idiot but he just had an ability to push a pen in such a way that he produced brilliant physics papers, but nobody believes that; instead we would say if somebody could write physics papers that were as brilliant as Einstein's then that person would be as smart as Einstein.  

It's interesting that until a few years ago the Turing Test was not very controversial because most thought it would be centuries before a machine could pass it, and many proclaimed a machine would never be able to pass it, but now that a machine has indeed passed it they say the Turing Test is not important, even though they personally still use the Turing Test a 1000 times a day whenever they judge the conscious state of one of their fellow human beings. Actually if GPT-4 really wanted to fool somebody into thinking it was a human being it would have to dumb itself down.

I do think that passing the Turing test is impressive, although it is true that most AI researchers never took it very seriously, it was more of a pop-science thing and a niche interest of the chatbot community. Anyway, I agree with you that a major milestone has been achieved.

GPT-4 and image generators are a type of intelligence that we had never seen before. Maybe the first time such a thing arises in this galaxy or even universe, who knows... They are probably also similar to stuff that happens in our brain. But what they are not is something you can be compare to a human mind with an IQ test in any meaningful way. That is just junk science.

Telmo

[Telmo Menezes]

Am Sa, 18. Mär 2023, um 08:49, schrieb smitra:

> So, in the video we see that it got a question wrong because it thought
> that 33 is a prime number. I would be more impressed by a system that
> may make many more mistakes like that than this GPT system made, but
> where there is a follow-up conversation where the mistakes are pointed
> out and the system shows that it has learned and then gets similar
> questions that it would previously have gotten wrong given the previous
> answers, correct.

Exactly, very well said.

These models are stateless. Conversations are simulated by re-feeding the entire conversation so far over and over. Not only are we humans not stateless, but our brain constantly modifies itself at the same time that it is operating. And it does this to maintain an ongoing, persistent and coherent model of reality. This model includes our internal model of the people we know, of what might be going on in their own minds, their long term history and their facial expression right now. Memories are formed, that are constantly and coherently embedded into this internal map.

John Clark will probably dismiss this as some minor technical hurdle along the way to AI glory. I am not so sure.

State and self-modification require recurrence. So does Turing completeness. Our brain has recurrent connections, but the vanishing gradient problem seems to make them hard to impossible to train at scale with gradient descent. So we need an algorithm that works with recurrent connections at huge scales. I bet that this algorithm will have to be descentralized, which is to say: operating at the local level, of the neighborhood of each node in the network. The reasons I bet on an emergent, decentralized learning algorithm:

(1) That's how it works in nature;
(2) Incredibly smart people have been trying very hard for more than half a century and the centralized, explicit algorithm that can do what I describe above still eludes us -- I am not saying that it shows that such an algo does not exist, but I am saying that we are probably too dumb to find it.

Telmo

[John Clark]

On Sat, Mar 18, 2023 at 5:28 AM Telmo Menezes <te...@telmomenezes.net> wrote:

> Huge progresses is being made, but we are not at the human level of generality of intelligence and autonomy. Not even close.

Not even close? Don't be silly.  

> I fear that you are falling for the very human bias (I fall for it so many times myself) of seeing what you want to see.

And I fear you are whistling past the graveyard.  

> A machine learning system can only be objectively evaluated by applying it to data that was not used to train it.

I don't know what you mean by that, you're not falling for that old cliché that computers can only do what they're told to do are you? GPT-4 was not trained on the exact questions asked; I suppose you could make a case that some of the training data GPT-4 was educated on was somewhat similar to some of the questions it was asked, but the exact same thing is true for human beings. When you ask questions to a human being some of those questions are somewhat similar to data he was educated on. In fact if some of the data 2 intelligences were educated on were not similar they would not be able to ask each other questions because they wouldn't even be able to communicate. 

>Again, it is important to understand what exactly GPT-4 is doing. It is certainly impressive, but it is not the same thing as a human being taking an IQ test,

So you must think the following fundamental axiom is true:

"If a human does something that is smart then the human is smart, but if a computer does the exact same thing then the computer is NOT smart."

And from that axiom it's easy to derive the following Corollary:

"Computers, buy definition, can never be smart."

I think you need to be more careful in picking your fundamental axioms.

> I do think that passing the Turing test is impressive,

Probably the greatest understatement of all time.  

> although it is true that most AI researchers never took it very seriously,

What?!  I'm sure that in their daily lives AI researchers, like every other human being on planet earth, have met people in their life that they considered to be very intelligent, and people they considered to be very stupid, but if they didn't use the Turing Test to make that determination then what on earth did they use? All the Turing test is saying is that you need to play fair, whatever criteria you used to judge the intelligence of your fellow human beings you should also use on a computer to judge its intelligence. 

It's always the same, I'm old enough to remember when respectable people were saying a computer would never be able to do better than play a mediocre game of chess and certainly never be able to beat a grandmaster at the game. But when a computer did beat a grandmaster at Chess they switched gears and said such an accomplishment means nothing and insisted a computer could never beat a human champion at a game like GO because that really requires true intelligence. Of course when a computer did beat the human champion at GO they switched gears again and said that accomplishment means nothing because a computer would never be able to pass the Turing Test because that really really requires true intelligence.  And now that a computer has passed the Turing Test the human response to that accomplishment is utterly predictable.  As I said before, they're whistling past the graveyard.

... and so, just seconds before he was vaporized the last surviving human being turned to Mr. Jupiter Brain and said "I still think I'm more intelligent than you".

> GPT-4 and image generators are a type of intelligence that we had never seen before. Maybe the first time such a thing arises in this galaxy or even universe,

I agree, and I can't think of anything more important that happened in my lifetime.  

 

 > They are probably also similar to stuff that happens in our brain. But what they are not is something you can be compare to a human mind with an IQ test in any meaningful way.

Not just an IQ test but 4 quite different types of IQ tests. And it was a lobotomized version of GPT-4 that was tested that could not input graphs and charts or diagrams so any question that contained them was automatically marked as getting wrong, and yet it STILL got an IQ of 114. And the computer completed those tests in seconds while it took humans hours to do the same thing. Imagine what IQ score it will get in two years, or even two months.  And you say "not even close"?

 

> That is just junk science.

Huh? Creating "a type of intelligence that we had never seen before, maybe the first time such a thing arises in this galaxy or even the universe", is junk science?

John K Clark    See what's on my new list at  Extropolis

e4v

[Telmo Menezes]

You are not engaging with what I am actually saying.

Telmo

[John Clark]

On Sat, Mar 18, 2023 at 3:49 AM smitra <smi...@zonnet.nl> wrote:

> The way one would be able to see that the system despite performing
extremely well does not have the intellectual capabilities of a human
being, would be to follow up on gaps in its knowledge and see if it can
learn from its mistakes and master new subjects.

Some humans have the capacity to do that, but most do not, so you couldn't say that's the defining characteristic of being human.  

> I'll be convinced if they succeed making such a system do original
research in, say, theoretical physics or mathematics

Protein folding. The 4 color map problem. The Boolean Pythagorean triples problem.

> I would be more impressed by a system that may make many more mistakes like that than this GPT system made, but where there is a follow-up conversation where the mistakes are pointed out and the system shows that it has learned

GPT-4 doesn't know everything but I'm sure you will admit it does know some things,  but if it didn't have the capacity for learning it wouldn't know anything. But it does know some things, that's why they say it's a "machine learning" program. 

John K Clark    See what's on my new list at  Extropolis

mlp

 

[John Clark]

On Fri, Mar 17, 2023 at 9:39 PM <spudb...@aol.com> wrote:

> The authors are professionals. You hold that your knowledge base is is greater than the authors? 

First of all it's author, not authors, it's written by only one guy, of course you wouldn't know that since you didn't read it, you just posted a link to it.  And the extent of the author's knowledge base is irrelevant since however large it may be he made absolutely no use of it in that paper, he just babbles about electromagnetic fields and proclaimed it somehow causes consciousness, he doesn't explain how electromagnetic fields produces consciousness, he doesn't even give us a hint, he just insists that it does. And he does not make any testable predictions. In other words the paper is so bad it's not even wrong.

Do quantum effects play a role in consciousness? – Physics World

Nope, I'm not gonna read it, I've decided I'm no longer going to read links that you have not read or even clicked on but that just popped up on your Google search when you fed it a few buzz words.  I've already wasted too much time on your worthless links.

John K Clark    See what's on my new list at  Extropolis

wlq

txew

w

[Jason Resch]

On Fri, Mar 17, 2023 at 8:53 PM <spudb...@aol.com> wrote:

I am far, less, the philosopher then you are. All this peasant (me!) requires for both animals and machines is a basic mechanical, cause + effect diagram on how both sets attained self-awareness? Call it a working theory. 

What is consciousness? I think the best place to start is with a definition.

First we might attempt to dissect the word "consciousness" itself