Path: utzoo!attcan!uunet!nfsun!ditka!formtek!darth!pitt!cadre!geb
From: geb@cadre.dsl.PITTSBURGH.EDU (Gordon E. Banks)
Newsgroups: comp.ai
Subject: Re: Free Will & Self Awareness
Message-ID: <1209@cadre.dsl.PITTSBURGH.EDU>
Date: 23 May 88 13:28:44 GMT
References: <770@onion.cs.reading.ac.uk> <1177@bingvaxu.cc.binghamton.edu> <10942@sunybcs.UUCP> <31024@linus.UUCP> <1176@cadre.dsl.PITTSBURGH.EDU> <1187@crete.cs.glasgow.ac.uk>
Reply-To: geb@cadre.dsl.pittsburgh.edu.UUCP (Gordon E. Banks)
Distribution: comp
Organization: Decision Systems Lab., Univ. of Pittsburgh, PA.
Lines: 15

In article <1187@crete.cs.glasgow.ac.uk> gilbert@cs.glasgow.ac.uk (Gilbert Cockton) writes:
>In article <1176@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu.UUCP (Gordon E. Banks) writes:
>>Punishment can serve to "redesign" the human machine.  If you have children, 
>>you will probably know this.  Unfortunately, it doesn't work with everyone.
>How can an artificial intelligence ever rise above recidivism?  Are
>there any serious examples of re-programming systems, i.e. a system
>that redesigns itself in response to punishment.

Certainly!  The back-propagation connectionist systems are "punished"
for giving an incorrect response to their input by having the weight
strengths leading to the wrong answer decreased.  In most connectionist
learning systems such "punishment" is used exclusively, not rewards.  
You may consider this a "crude" learning system, but it probably isn't
much cruder than that actual neural apparatus underlying most organic
brains.