[comp.ai.philosophy] A New Turing Test!!!

thomas@ckgp.UUCP (Michael Thomas) (06/27/91)

In article <1991Jun25.180330.3645@cs.yale.edu>, ciancarini-paolo@cs.yale.edu (paolo ciancarini) writes:
> In article <612@ckgp.UUCP> thomas@ckgp.UUCP (Michael Thomas) writes:
> >Hi everyone,
> >        
> > Has anyone established a test or heard of one that would be better suited
> > for deturmining if a system is intelligent?
-------------------------
Ahhh!!! - I did it again, I'm sure you understand what I mean, but the
real question is:

	Has anyone established a test or heard of a test that would
be better suited for deturmining if an AI system meets up to the 
standards of the AI community?
-------------------------
> 
>Recently I saw on some TV channel a documentary about sen. McCarthy
>and his times (the late forties and early fifties). 
>I was struck by a question made by a journalist to the senator: 
>McCarthy was asked how he could recognize a communist.
>He answered something like this: 
>"if he speaks like a communist, if he writes like a communist,
>if he has communists friends, if he looks like a communist, he IS a communist".
>Unfortunately, the Senator could convince a lot of people
>that such a criterion was good.

	This is a good point, and That is my main concern. Are we
not looking at the question of AI correctly, with a foundation 
from the 50's. Alot has changed about our understanding of people
and the brain and the mind. This directly affects the way we must
direct and focus our views in AI.
	I also hold a strong position in that "intelligence"
(as evil as that word is) can be aquired/functional in different
forms. I might have mentioned this before, but human language
is based largely on words. I have heard conjecture, that dolphins
language is largely based on actual images, which they can produce
the images with clicks and send them over vast distances. The point
of this is that yes there are different forms of "human" intelligence
(math,english,compsci,etc) but there are still different forms of
intelligence like (maybe) one of dolphins. What I'm getting at is that
a computer might be more "intelligent" (in reference to maintainable
knowledge) and still not behave like a human or respond like one.
one thing that comes to my mind is that if I had everything I knew
on-line with easy accessablity, and could be almost possitive that the
information was correct, and understand more that 14,000 words
(adult vocab:12,000 - 14,000words) I am sure that this would be a
jump in several aspects of mental performance in general. (given
a good method of using the info...) To be able to compare and contrast
thousands of variables and do stitistic on them and, well do things
that I just don't do with every day information. maybe a better understanding
of chaos sciences (everything happens in cycles, specifically strange
cycles) [I'm sure you all get the idea, that it will be different, no
matter what...]
	So back to the original question: how can we establish a
test or way of deturmining if a system lives up to that of what
is expected by the AI comunity as a whole?

thanks!   8^)     [so, what does everyone think?]
-- 
Thank you,
Michael Thomas
(..uunet!ckgp!thomas)

logan@nexus.yorku.ca (Beryl Logan) (06/27/91)

thomas@ckgp.UUCP (Michael Thomas) writes:

|...'intelligence'
|(as evil as that word is) can be aquired/functional in different
|forms. I might have mentioned this before, but human language
|is based largely on words. I have heard conjecture, that dolphins
|language is largely based on actual images, which they can produce
|the images with clicks and send them over vast distances. The point
|of this is that yes there are different forms of "human" intelligence
|(math,english,compsci,etc) but there are still different forms of
|intelligence like (maybe) one of dolphins. What I'm getting at is that
|a computer might be more "intelligent" (in reference to maintainable
|knowledge) and still not behave like a human or respond like one.

It seems here you are conflating intelligence with language.  That
there is language is a criterion for the claim that there is
intelligence, a language being the sort of thing with rules, syntax,
etc.  There are different forms of language, things that have these
features, such as `math,english,compsci,etc' and, if we knew about its
rules, syntax, etc. `dolphinese' which are criteria, are what gives us
license to say, that an organism is intelligent.