From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!csd.unb.ca!morgan.ucs.mun.ca!nstn.ns.ca!aunro!alberta!ubc-cs!uw-beaver!micro-heart-of-gold.mit.edu!mintaka.lcs.mit.edu!olivea!spool.mu.edu!wupost!uwm.edu!linac!convex!constel Thu Feb 20 15:21:49 EST 1992
Article 3839 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!csd.unb.ca!morgan.ucs.mun.ca!nstn.ns.ca!aunro!alberta!ubc-cs!uw-beaver!micro-heart-of-gold.mit.edu!mintaka.lcs.mit.edu!olivea!spool.mu.edu!wupost!uwm.edu!linac!convex!constel
lation!a.cs.okstate.edu!onstott
>From: onstott@a.cs.okstate.edu (ONSTOTT CHARLES OR)
Newsgroups: comp.ai.philosophy,sci.tech.philosophy,talk.misc.philosophy
Subject: Aristotelian Ontology and AI
Message-ID: <1992Feb18.195936.14369@a.cs.okstate.edu>
Date: 18 Feb 92 19:59:36 GMT
Organization: Oklahoma State University, Computer Science, Stillwater
Lines: 89


	I thought I might present a break from the typical logico-
semantic argumentation on here by talking about Aristotelian
Ontology in regards to Artificial Intelligence.
 
 	I am going to present a particular view of Artificial
Intelligence, and this is not taken to be THE view of AI, and
analyze it under the precepts of an Aristotelian ontology.

	There exists a view of Artificial Intelligence that living
things are essentially reducible to predictable material events. 
It is also thought that these events can be mapped onto a
computer.
In short, we can reproduce (human intelligence), by either: 1.)
Reproducing an exact (cognitive structure) on a computer thus
reproducing the biological (cognitive structure) in a binary
physically determined system resulting in behavior identical to a
behaving agent, or 2.) Reproduce a (cognitive structure) onto a
computer which results in an actual reproduction of the person. 
I put some of these things in parenthesis to indicate that these
terms are not exactly clear as to their meaning.  

	An argument was presented, by analogy of a computer program,
which said that if we hold that we can reproduce the cognitive
structure onto a computer, then we have, in fact, reproduced that
person. If we apply something like the Aristotelian abstraction,
this conclusion can be considered correct because "knowledge,"
"skin color," "physical composition," etc can be considered
"accidentals" and are not necessarily relevant to the thing that
it is.  However, this immediately takes us into the problem of
the Universal versus the Particular, where I have argued
elsewhere(Onstott, Charles, "The Universal as an Epistemological
Category"), that "the Universal is an epistemological category
and that the particular is an ontological category."   Further,
"The Universal is achieved through the particulars(which
ontologically are dialectically prior) but the Being of the
particulars remains unknown without the Universal(which
epistemologically is dialectically prior)."  In this way, then,
any reproduction of a program, or a person on a computer for that
matter, is not ontologically the same but is epistemologically
the same.  The idea is in level of degree:  In reality we can not
say that a program is exactly the same when it runs on different
physical machines(the machines, at least, are different); all
that we can say is that the work that it does; or the outputs
that it gives are, in fact, the same.  However, we identify and
label the program what it is by virtue of what it does, the
"workly" element(Heidegger, Being and Time, The Origin of the
Work of Art, etc.) of the program.  Therefore, a the statement "A
reproduction of a cognitive structure necessarily leads to a
reproduction of the person" can only be said with due
qualification.

	Its impact on Artificial Intelligence as a whole is probably
relevant as well.  Instead of arguing in circles about the
problem of "Strong AI," I think it can be dissolved from an
Aristotelian ontology.  Strong AI, as well as many identity
theories, fails mainly because there would be no such thing as
exact reproduction.  Strong-AI must remain purely externalistic
in evaluation and can not make any claims about the actual
physical structure of a human being.  Further, any research
program that attempts to define a human being in terms of
Ontology also fails because nothing ontologically significant has
been said.  However, it does appear that an argument can be made
on an epistemological ground.  If it acts like a human being,
then it is a human being('is' in this sense refers to the
epistemology not the ontology).  In this way, we have a purely
externalistic approach to understanding what a human might be. 
Then, AI simply becomes a methodological approach to abstracting
the essence of human beings; but in no way provides an
ontological answer.  At this point, however, I am afraid we will
go on for ever and ever because I am not certain, perhaps due to
ignorance, that there is an essential essence to be found.  

	From this view, the science of Artificial Intelligence makes
a lot more sense to me(yes, I actually think maybe I can find
justification now**Hurrah***).  However, the science must be
careful not to make ontological statements about human beings as
in this sense the act of science is not to recreate or, if you
will, play God; but simply, like a good accountant, to report.

------------------------------------------------------------------------
Charles O. Onstott, III                  P.O. Box 2386
Undergraduate in Philosophy              Stillwater, Ok  74076
Oklahoma State University                onstott@a.cs.okstate.edu

"The most abstract system of philosophy is, in its method and purpose, 
nothing more than an extremely ingenious combination of natural sounds."
                                              -- Carl G. Jung
-----------------------------------------------------------------------


