Newsgroups: comp.ai,comp.ai.philosophy,sci.logic,sci.cognitive
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!news.sprintlink.net!howland.reston.ans.net!agate!news.ucdavis.edu!csus.edu!netcom.com!jqb
From: jqb@netcom.com (Jim Balter)
Subject: Re: FIRST order?
Message-ID: <jqbDBspDn.3C9@netcom.com>
Organization: NETCOM On-line Communication Services (408 261-4700 guest)
References: <jqbDBpCx1.o5@netcom.com> <3u78iv$pvd@saba.info.ucla.edu> <BILL.95Jul15134709@cortex.nsma.arizona.edu> <3u9vaq$84o@saba.info.ucla.edu>
Date: Sun, 16 Jul 1995 06:15:23 GMT
Lines: 46
Sender: jqb@netcom7.netcom.com
Xref: glinda.oz.cs.cmu.edu comp.ai:31576 comp.ai.philosophy:30285 sci.logic:12499 sci.cognitive:8337

In article <3u9vaq$84o@saba.info.ucla.edu>,
Michael Zeleny <zeleny@oak.math.ucla.edu> wrote:
>bill@nsma.arizona.edu (Bill Skaggs) writes:
>>zeleny@sepulveda.math.ucla.edu (Michael Zeleny) writes:
>
>>   > [ . . . ] But it is often depressing to see how little the AI
>>   > theoreticians understand the most basic accomplishments of
>>   > contemporary logic.  If the current crop is as representative of
>>   > its kind as McCarthy and Minsky were of the original artificial
>>   > intelligentsia, there is little reason to hope that this
>>   > situation will change any time soon. 
>
>>This is a tantalizing remark, but I'm not quite sure I understand you.
>>Suppose I'm an AI guy whose goals are purely behavioral:  I want to
>>make a machine that moves, talks, and solves problems like a human,
>>but I don't care what's going on inside it.  Does contemporary logic
>>have anything that will help me achieve this?
>
>I cannot help you with your desire to mimic human motion, but anyone
>interested in the production of human speech would have to take into
>account the advances in our understanding of syntax, associated with
>the development of transformational grammar, and militating against
>behavioristic methodology.  As regards problem-solving, one of the
>salient tasks is to characterize the class of inferences that humans
>are capable of making.  Two obvious tools suited for accomplishing
>this are general recursion theory and ordinal logics.  Other relevant
>developments include much of the recent work in model-theoretic logic
>and its philosophical implications (e.g. the question of the human
>ability to reason following the \omega-rule) and the study of various
>alethic, epistemic, and deontic modalities.  This list is entirely
>determined by the extent of my knowledge and the orientation of my
>research interests.  Perhaps others can suggest different topics.

Ah, perhaps we are getting somewhere.  What it would take to produce a robot
that could mimic a human being is overwhelming in its complexity, and puts to
shame the naive (and "RSN") predictions and expectations of many AI
researchers (I am not saying that all AI researchers put forth such
predictions or hold such expectations).  This catalogue could be extended
almost indefinitely, making the task seem even more daunting, and perhaps
scaring beginnners out of the field.  Of course, all these are *practical*
considerations, and have no direct bearing on the ethical and mathematical
arguments against Strong AI.  (Remember that comp.ai and comp.ai.philosophy
are different groups with different interests.)
-- 
<J Q B>

