Xref: utzoo comp.ai:2696 talk.philosophy.misc:1612
Path: utzoo!attcan!uunet!convex!killer!osu-cis!tut.cis.ohio-state.edu!bloom-beacon!gatech!linus!mbunix!bwk
From: bwk@mitre-bedford.ARPA (Barry W. Kort)
Newsgroups: comp.ai,talk.philosophy.misc
Subject: Re: Artificial Intelligence and Intelligence
Summary: The limits of AI are the frontiers of AI.
Keywords: Modal Logic, Intuitionist Logic, Model-Based Reasoning
Message-ID: <42021@linus.UUCP>
Date: 23 Nov 88 17:16:19 GMT
References: <484@soleil.UUCP> <88Nov15.170837est.707@neat.ai.toronto.edu>
Sender: news@linus.UUCP
Reply-To: bwk@mbunix (Kort)
Organization: The MITRE Corporation, Bedford, Mass.
Lines: 24

I enjoyed reading Brad Brown's comments regarding Turing-computability
and the limits of AI.

It seems to me that one of the most powerful forms of human reasoning
is reasoning by analogy, or model-based reasoning.  We use models,
metaphors, parables and analogies to transform problems from one
domain to another, thereby borrowing ideas and translating them to
novel situations.

Reasoning by analogy requires pattern matching with respect to the
structure of a knowledge base.  We look at the shape of the semantic
network or the shape of the tree and ask ourselves if we have encountered
a similarly structured knowledge base before.  Then, mutatis mutandis,
we map out the analogy and fill in the missing pieces.  Natural language
is a powerful tool for suggesting metaphors.

In mathematical circles, proof by analogy is emerging as an interesting
research frontier.  Saul Kripke has done seminal work in modal reasoning
and intuitionist logic, which formalize these ideas.

I wonder how the "limits of AI" argument will look after machines
learn how to mimic human-style model-based reasoning.

--Barry Kort