From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!psych.toronto.edu!michael Mon Mar  9 18:33:22 EST 1992
Article 4090 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
Message-ID: <1992Feb27.213832.25085@psych.toronto.edu>
Organization: Department of Psychology, University of Toronto
References: <1992Feb25.165326.16204@ccu.umanitoba.ca> <1992Feb25.190913.7908@psych.toronto.edu> <kimhock.84@csar.uucp>
Date: Thu, 27 Feb 1992 21:38:32 GMT

In article <kimhock.84@csar.uucp> kimhock@csar.uucp (Ng Kim Hock) writes:

>I still cling on to the opinion that part of the hard work of interpreting 
>the semantics can be redirected to the party interacting with the machine. 

Then the *machine* has no understanding, which is the question under
debate.

>Consider a researcher when he represents a physical phenoma into a 
>Mathematical system and then uses Mathematical rules to manipulate the 
>symbols then convert the research finding by interpreting it with regards 
>to the phenomena. 
>
>Imagine that the Mathematics part is taken over by a Machine and the 
>scientist interacts with the machine with the machine getting back to the 
>scientist with the result. To the scientist who need to do the job of 
>interpretating the result with respect to the phenomena, he can say that the 
>machine is intelligent because it is able to predict a correct result which 
>he himself would have taken arrived at.
>
>The machine does not understand the physical phenomena but it is able to 
>manipulate the symbolic representation in a consistent way which can be 
>interpreted into something sensible by an intelligent being. 
>
>Thus I disagree with the statement that the machine has "no chance" of 
>conversing intelligently with an intelligent being. The magic is to let the 
>machine specialise in symbolic manipulation and the interlocutor specialise 
>in interpretation.

This is fine, and corresponds precisely to the Chinese Room situation.  The
machine *can be interpreted* as understanding, but doesn't possess it.

- michael
 



