Newsgroups: sci.lang,sci.psychology,rec.arts.books,comp.ai.philosophy
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!oitnews.harvard.edu!news.sesqui.net!academ!bcm!cs.utexas.edu!utnut!utgpu!pindor
From: pindor@gpu.utcc.utoronto.ca (Andrzej Pindor)
Subject: Re: Chomsky on Consciousness and Dennett
Message-ID: <D9IA6y.73p@gpu.utcc.utoronto.ca>
Organization: UTCC Public Access
References: <3q2put$p07@Mercury.mcs.com> <3qa4dd$9rm@ixnews2.ix.netcom.com> <3qfncu$l0i@acmex.gatech.edu> <JMC.95May30183530@SAIL.Stanford.EDU>
Date: Thu, 1 Jun 1995 18:04:09 GMT
Lines: 55
Xref: glinda.oz.cs.cmu.edu sci.lang:39678 sci.psychology:42374 comp.ai.philosophy:28509

In article <JMC.95May30183530@SAIL.Stanford.EDU>,
John McCarthy <jmc@cs.Stanford.EDU> wrote:
>Tim Dodd includes:
>
>     If I build a simple robot which can successfully negotiate a
>     maze, then I would say that the robot "understands" how to
>     find its way through the maze.  I would say the same thing
>     of a human being who could solve the maze.  I see no reason
>     to have two different words for identical behavior simply
>     because one behaver is human and the other is not.
>
>I am not inclined to use the word "understands" for the simplest
>performances.  Suppose someone understands in Tim Dodd's sense how to
>drive a car.  Suppose now he comes out of his house, but the wheels
>have been removed from his car - as he and everyone can see.  With the
>simplest kind of understanding, he would still get into the car, start
>the engine, put the car in gear and step on the gas.  I was going to
>say he would expect the car to work, but expectation of what will
>happen is already more than Dodd has committed himself to.
>
I am sorry, but this argument is phoney. By the same reasoning one may claim
that no one who can't avoid a crash when running into a patch of black ice
(assumming a highly skilled driver could do it) "understands how to drive
a car", etc. "Understanding" would then be limited only to top experts in any 
field or to extremely simple phenomena.

>Of course, Dodd may argue if he wants to that the extreme doer
>nevertheless understands, but I hope he won't, because the interesting
>kinds of understanding involve quite a bit more than executing a
>program that performs some act.
>
And how do you determine what is this "interesting kind of inderstanding"?
It seems to be a very flexible notion, ideally suited to be used selectively,
when need arises, to discredit an oponent.
Of course, if you have an unambiguous criterion for deciding what is 
"interesting kind of understanding" then I am wrong and I applologize.
.........
>
>Some past and current AI programs have come closer to interesting
>understanding, but we don't even have a characterization of
>understanding yet.

Exactly, the only way we recognize understanding is by behavior. Bingo!

>-- 
>John McCarthy, Computer Science Department, Stanford, CA 94305
>*
>He who refuses to do arithmetic is doomed to talk nonsense.

Andrzej
-- 
Andrzej Pindor                        The foolish reject what they see and 
University of Toronto                 not what they think; the wise reject
Instructional and Research Computing  what they think and not what they see.
pindor@gpu.utcc.utoronto.ca                           Huang Po
