Path: utzoo!attcan!uunet!mcvax!ukc!strath-cs!glasgow!gilbert
From: gilbert@cs.glasgow.ac.uk (Gilbert Cockton)
Newsgroups: comp.ai
Subject: Re: letter to THE NEW YORK REVIEW concerning AI
Keywords: Intention
Message-ID: <2564@crete.cs.glasgow.ac.uk>
Date: 10 Mar 89 13:25:40 GMT
References: <7471@venera.isi.edu> <Feb.9.21.07.59.1989.2316@elbereth.rutgers.edu> <2447@crete.cs.glasgow.ac.uk> <509@mmlai.UUCP> <2481@crete.cs.glasgow.ac.uk> <2369@cadre.dsl.PITTSBURGH.EDU>
Reply-To: gilbert@cs.glasgow.ac.uk (Gilbert Cockton)
Organization: Comp Sci, Glasgow Univ, Scotland
Lines: 30

In article <2369@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu (Gordon E. Banks) writes:
>In article <2481@crete.cs.glasgow.ac.uk> gilbert@cs.glasgow.ac.uk (Gilbert Cockton) writes:
>The system can be said to "understand" how to do something
>(say, a robot that spots for a gymnast) when it performs correctly under 
>supervision by someone who "understands" correct performance.

Agreed, almost.  At this stage, the supervisor "trusts" the robot.  It
has to demonstrate some period of continued competence before it would
be accepted as having a good understanding.

At this point, most problems do not require further "training", just a
comment.  As far as I know, there is only one way to "train" a neural
network, whereas the growth of understanding in the presence of an
expert passes through several forms of training, which may eventually
reduce to the shake of a head and a gesture.

Is there a sense in which neural network training requires

	a) an artificial, well-designed task 
	b) continued practice over this task.

If so, then ho, ho, ho, because life just ain't so simple.  The good
guys in this world are the ones who don't need Skinnerian programmed
instruction.

The topic is life, not the behavioural modification of the mentally ill
or deficient.
-- 
Gilbert Cockton, Department of Computing Science,  The University, Glasgow
	gilbert@uk.ac.glasgow.cs <europe>!ukc!glasgow!gilbert