From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!psych.toronto.edu!michael Tue Mar 24 09:54:54 EST 1992
Article 4394 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: Definition of understanding
Organization: Department of Psychology, University of Toronto
References: <1992Mar10.150226.14196@gpu.utcs.utoronto.ca> <1992Mar10.171111.6954@psych.toronto.edu> <1992Mar10.210329.1904@gpu.utcs.utoronto.ca>
Message-ID: <1992Mar11.165622.19854@psych.toronto.edu>
Date: Wed, 11 Mar 1992 16:56:22 GMT

In article <1992Mar10.210329.1904@gpu.utcs.utoronto.ca> pindor@gpu.utcs.utoronto.ca (Andrzej Pindor) writes:

>As I have pointed out in a reply to your another posting, if you put things 
>this way, then you have to allow for the database on Chinese symbols to have 
>comparable content to your database on Engilish words. You will agree that in
>principle to answer the question to the story in English (has the man eaten 
>the hamburger?) you do not need, for instance, to know what hamburger looks like
>or what it tastes like (it is enough to know that a burned hamburger is a no-no)
>The database for a SHRDLU program (and the CR) has only the info required to
>answer the questions of the type posed.

But the questions that can be posed to the CR are *any* question, including
"What does a hamburger look like?  What does a hamburger taste like?"  By 
assumption, it answers such questions correctly.  But I would agree that
the system doesn't know what a hamburger tastes or looks like. 

It is an assumption of Strong AI that *any* information can be encoded
into the database, including the look and taste of a hamburger.  If you
disagree with this, then the project of Strong AI is doomed from the
start.

I suggest we take our discussion to email, since we seem to be responding
back and forth without input from other folks.

- michael
 
 


