From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!psych.toronto.edu!michael Mon Mar  9 18:35:56 EST 1992
Article 4328 of comp.ai.philosophy:
Newsgroups: comp.ai.philosophy
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!psych.toronto.edu!michael
>From: michael@psych.toronto.edu (Michael Gemar)
Subject: Re: Monkey Room
Organization: Department of Psychology, University of Toronto
References: <1992Mar5.145435.11897@gpu.utcs.utoronto.ca> <1992Mar5.200829.708@psych.toronto.edu> <1992Mar6.142314.12046@gpu.utcs.utoronto.ca>
Message-ID: <1992Mar6.220046.21132@psych.toronto.edu>
Date: Fri, 6 Mar 1992 22:00:46 GMT

In article <1992Mar6.142314.12046@gpu.utcs.utoronto.ca> pindor@gpu.utcs.utoronto.ca (Andrzej Pindor) writes:
>In article <1992Mar5.200829.708@psych.toronto.edu> michael@psych.toronto.edu (Michael Gemar) writes:
>>In article <1992Mar5.145435.11897@gpu.utcs.utoronto.ca> pindor@gpu.utcs.utoronto.ca (Andrzej Pindor) writes:
>>>In article <1992Mar4.210902.28435@psych.toronto.edu> michael@psych.toronto.edu (Michael Gemar) writes:
>>>>
>>>>Well, it shows that the Turing Test is not infallible.  This in itself is
>>>>a useful reminder for folks here.  It also shows that arguments against
>>>>the "Turing Test" results that were posted here a while ago, in which
>>>>some laypeople thought programs were actually computers, are at best
>>>>ad hoc.  There is no *clear* way to conduct a Turing Test, and no way
>>>>that will yield perfect results.
>>>>
>>>You must be very young if you still look for perfection. Prepare yourself for
>>>a big shock :-).
>>>(Sorry, couldn't restrain myself)
>>
>>Well, Andrzej, the AI community must be downright infantile, since its members
>>continue to mouth the words "Turing Test" as if it were some kind of magical
>>incantation that simply *proves* that a certain system is intelligent.
>>
>>- michael
>>
>>(and yes, I *am* getting snarky...)
>>
>Michael, you know the AI comunity certainly better than I do. So tell me:
>did any serious member of this community ever claimed that TT will yield
>*perfect* results? And do you know any more reliable test for intelligence of
>a system?  I am sure they would be happy if you proposed some other, better
>test. Until there is one, they are using what is available. What would you
>expect them to do? To say "There is no fool-proof test for intelligence at 
>present, so we might just as well close the shop"? In any science, is there ever
>a perfect test? Look for instance at medical sciences. They are full of claims,
>counterclaims, often allegedly identical experiments giving completly different
>results in different labs. And how about cold fusion controversy? Does this
>mean that all this sciences are bogus and should be discontinued?

As I noted in another posting, I am amazed that my off-hand remarks have
raised such ire.  My point was merely that passing the TT is not a sufficient
condition for demonstrating intelligence.  Yes, I realize that scientific
tests in general are not perfect.  Yes, I realize that the TT is probably
pretty good in those cases that we want to look at (avoiding the CR
case, for now, which brings the whole thing into question again).  All
I was pointing out is that it isn't perfect.  If everyone accepts that, fine,
there was no need for my comments.

BTW, I am intrigued that no one has pursued my remarks as to what exactly
*does* constitute the TT...


- michael




