From newshub.ccs.yorku.ca!torn!utcsri!rutgers!uwm.edu!wupost!usc!elroy.jpl.nasa.gov!swrinde!gatech!purdue!mentor.cc.purdue.edu!f3w Tue Jul 28 09:41:39 EDT 1992
Article 6478 of comp.ai.philosophy:
Xref: newshub.ccs.yorku.ca rec.arts.sf.science:2788 comp.ai.philosophy:6478
Path: newshub.ccs.yorku.ca!torn!utcsri!rutgers!uwm.edu!wupost!usc!elroy.jpl.nasa.gov!swrinde!gatech!purdue!mentor.cc.purdue.edu!f3w
>From: f3w@mentor.cc.purdue.edu (Mark Gellis)
Newsgroups: rec.arts.sf.science,comp.ai.philosophy
Subject: Re: How do computers fare on scholastic achievement tests?
Message-ID: <54357@mentor.cc.purdue.edu>
Date: 17 Jul 92 17:59:32 GMT
References: <1992Jul14.014658.4921@newstand.syr.edu> <NICKH.92Jul14141610@VOILA.VENARI.CS.CMU.EDU> <1992Jul16.093057.8880@techbook.com>
Followup-To: rec.arts.sf.science
Organization: Purdue University Computing Center
Lines: 48


I think we may be straying from the real issue.  Computers can certainly
outdo humans in a lot of areas that be described as "intelligence."  In fact,
that is why we built them.  If you think of intelligence simply as problem-
solving, then computers are ideal for certain kinds of intelligence without
ever having to be intelligent in other ways.  

Right now, of course, computers generally solve problems in an entirely
different way than the way humans think.  They run through a problem by 
making it considering all the reasonable (and sometimes just all) the
possibilities and then running through them at ultrahigh speed.  That is
why a computer can beat you at chess, it can simply ask itself "What are
all the possibilities for the next four moves that can be made and what
will leave me in the best situation, in terms of chess points" and it
chooses the one that works best.  And it can do this in a few seconds.
A human chess player, as I understand it, recognizes patterns and 
strategies, rather than blindly considering every possible combination
of moves.

Let me give you another example.  I have a program that will let computers
write poetry.  Some of the poems are even pretty good, in that haunting way
that sudden strange (i.e., random) twists of lyric are.  But can any computer
tell you why a poem is good or bad?

Even if you could make a critic out of a computer, you may still be leaving
out what some people call the "real" issue.  Self-awareness.  Of course, it
is hard to say where that kind of awareness begins.  Is a dog self-aware?
You bet it is.  Is a mouse or a bird?  Probably.  What about a praying mantis?
It hunts; it does not blindly follow pheremone instructions.  But is it aware
of itself as a being?  What about ants or rotifers?

Without falling back on some romantic notion of "soul," I would say that a
really good test of intelligence/self-awareness would be to have a computer
that could write a poem or story, or create a painting, and then tell you
why it did certain things in that effort, and why it is a good or bad piece.
This is not simply because these are creative efforts, but also because they
are communicative efforts (if you are trying to please someone in a certain
way with a painting, you are communicating with them) and communication 
of this sort requires that the creator is aware of itself in relationship
to other beings.   

Anyone know of good stories, by the way, where you have a "low" machine 
intelligence?  The machine is awake and aware, and can do some things a
trillion times faster than people, and it can understand spoken instructions,
very complex ones, but in terms of a being it is only about as smart as,
say, a Welsh terrier?

 


