Besides Turing, what other tests have been proposed for AIs?
Scott T. Jensen
determining if AIs have an intelligence equal to or greater than humans?
Scott Jensen
--
Like a cure for A.I.D.S., Alzheimer, Parkinson, & Mad Cow Disease?
Volunteer your computer for folding-protein research for when it's idle.
Go to http://www.distributedfolding.org/ to sign up your computer.
rick++
Could Androids completely pass as humans? Some androids even thought they were
human, as we wonder about the Ford character at the end of the film.
Supposedly an android lacked sophisticated "emotional intelligence",
which a skilled interrogator could distinguish.
The first Star Trek series explored these themes several times too.
rick++
was an intelligent activity. However, by brute force, the top computer can
been 99.9% of human beings.
Also, there might some problem communicating with an intelligence vastly
smarter than a human being. They much have too complex communication system
for humans to master, or think thoughts and have experiences that are too
high over our heads to grasp. An example are the aliens in the movie Contact
who have to use a kind of "cosmic baby talk" to communicate with humans.
What kind of conversation would you have with an ant?
Scott T. Jensen
I am more interested in non-fiction ones. Ones actually proposed by
scientists as a good test for general intelligence in an artifact. Or
possibly some significant milestone that if a computer program were to
achieve or surpass, a scientist(s) would say that it possess human-level or
great intelligence.
Pascal Bourguignon
"Scott T. Jensen" <s...@charter.net> writes:
> I am more interested in non-fiction ones. Ones actually proposed by
> scientists as a good test for general intelligence in an artifact. Or
> possibly some significant milestone that if a computer program were to
> achieve or surpass, a scientist(s) would say that it possess human-level or
> great intelligence.
I've got no references, but ISTR that some QI tests were passed
succesfully by some specific AI software.
The problem of these QI tests is that they may show intelligence only
if they're "suprizing", but once you've done a couple, they become
quite previsible (you can even train to increase your QI tests performance!)
That's why the Turing test is insurpassable: it's so loosely defined
that you can really convice yourself of the intelligence of the tested
subject. For example, it gives no time limit: you could easily spend
a week or a month on a Turing test!
--
__Pascal Bourguignon__ http://www.informatimago.com/
Nobody can fix the economy. Nobody can be trusted with their finger
on the button. Nobody's perfect. VOTE FOR NOBODY.
rick++
> scientists as a good test for general intelligence in an artifact. Or
> possibly some significant milestone that if a computer program were to
> achieve or surpass, a scientist(s) would say that it possess human-level or
> great intelligence.
Its because philosophers and science fiction writer imagine the "big picture"
while most scientists take timid little steps toward their next grant and paper.
j q st martin
> determining if AIs have an intelligence equal to or greater than humans?
"A CAPTCHA is a program that can generate and grade tests that:
Most humans can pass.
Current computer programs can't pass."
Gordon Joly
But ... "Intelligence is in the mind of the beholder" ...
Surely?
Gordo
--
This is not my sig nature.....
gordon joly @ The Internet....
Eray Ozkural exa
> Besides Alan Turing's test, what other tests have been proposed for
> determining if AIs have an intelligence equal to or greater than humans?
>
There is an online test called Longley-Sizemore test on
comp.ai.philosophy newsgroup. It is also a perpetual torture generator
process for those who already past it.
Cheers,
--
Eray Ozkural
Eray Ozkural exa
> "rick++" <ric...@hotmail.com> wrote:
> > The classic movie "Blade Runner" was a sophisticated
> > version of the Turing Test. Could Androids completely
> > pass as humans? Some androids even thought they
> > were human, as we wonder about the Ford character
> > at the end of the film. Supposedly an android lacked
> > sophisticated "emotional intelligence", which a skilled
> > interrogator could distinguish.
> >
> > The first Star Trek series explored these themes
> > several times too.
>
> I am more interested in non-fiction ones. Ones actually proposed by
> scientists as a good test for general intelligence in an artifact. Or
> possibly some significant milestone that if a computer program were to
> achieve or surpass, a scientist(s) would say that it possess human-level or
> great intelligence.
Yes, but maybe Voigt-Kampff test had a point!
John Haugeland concludes his "Authentic Intentionality" paper with the
following words:
"... cognitive science and artificial intelligence cannot succeed in
their own essential aims unless and until they can understand and/or
implement genuine freedom and the capacity to love."
If that is really the case, the true Strong AI test must detect the
capacity for love. How would you do that?
Regards,
--
Eray Ozkural