Path: utzoo!utgpu!water!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!bloom-beacon!AI.AI.MIT.EDU!NICK
From: NICK@AI.AI.MIT.EDU (Nick Papadakis)
Newsgroups: comp.ai.digest
Subject: [DEFRANCO@RADC-TOPS20.ARPA: Free Will]
Message-ID: <19880528032116.7.NICK@MACH.AI.MIT.EDU>
Date: 28 May 88 03:21:00 GMT
Sender: daemon@bloom-beacon.MIT.EDU
Organization: The Internet
Lines: 56
Approved: ailist@ai.ai.mit.edu

Date: Thu, 26 May 88 14:16 EDT
From: Carl DeFranco <DEFRANCO@RADC-TOPS20.ARPA>
Subject: Free Will
To: ailist@AI.AI.MIT.EDU


       The long standing discussion of Free Will has left me a little
       confused.  I had always considered Free Will a subject for the
       philosophers and theologians, as in their knowledge domain,
       the concept allows attribution of responsibility to
       individuals for their actions.  Thus, theologians may discuss
       the concept of Sin as being a freely chosen action contrary to
       some particular moral code.  Without Free Will, there is no
       sin, and no responsibility for action, since the alternatives
       are total determinism, total randomness, or some mix of the
       two to allow for entropy.

       For my two cents, I have accepted Free Will as the ability to
       make a choice between two (or possibily more) available
       courses of action.  This precludes such silly notions as the
       will to defy gravity.  Free will applies to choice, not to
       plausibility.  These choices will be guided by the indivdual's
       experience and environment, a knowledge base if you would,
       that provide some basis for evaluating the choices and the
       consequences of choosing.  To the extent that an individual
       has been trained toward a particular behavior pattern, his/her
       choices may be predicted with some probability.  In other
       circumstances, where there is no experience base or training,
       choices will appear to be more random.

       In general, people do what they please OR what pleases them.
       It is this background guidance that changes from time to time,
       and inserts the mathematical randomity into whatever model
       used to predict behavior.  Today it may please me to follow
       the standard way of thinking in exploring some concept.
       Tomorrow I may be more pleased to head off in some oddball
       direction.  It is this "Free Will" choice, in my view, that
       creates the intelligence in human beings.  We may take in
       information and examine it from several points of view,
       selecting a course of action from those points, and adding the
       results to our experience., i.e. we learn.

       As I follow AI from the sideline in my job, I won't presume to
       prescribe The Answer, but it would appear that true Artificial
       Intelligence can be given to a computer when:
            1.  It can learn from its experience.
            2.  It can test a "What If" from its knowledge.
            3.  There is some limited range of allowable random
                selection.
       Perhaps I take a simplistic view, but there appear to be a
       number of one-sided viewpoints, either philosophic or
       technical.

       Carl DeFranco
       defranco@radc-tops20.arpa
-------