Newsgroups: comp.ai
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!news.mathworks.com!news.alpha.net!uwm.edu!vixen.cso.uiuc.edu!howland.reston.ans.net!ix.netcom.com!netcom.com!nagle
From: nagle@netcom.com (John Nagle)
Subject: Re: What would IT TAKE to HAVE a TRUE leader in AI?
Message-ID: <nagleD41xC0.Kq8@netcom.com>
Organization: NETCOM On-line Communication Services (408 261-4700 guest)
References: <3h6pf7$a8i@rs10.tcs.tulane.edu> <3hnu20$ogs@hawk.ee.port.ac.uk> <nagleD3z334.2p5@netcom.com> <3hqknh$eaj@srvr1.engin.umich.edu>
Date: Wed, 15 Feb 1995 17:16:48 GMT
Lines: 16
Sender: nagle@netcom17.netcom.com

indigo@engin.umich.edu (Glenn E Taylor ) writes:
>>     It's worth remembering occasionally that AI can't even do a good
>>lizard-level brain yet, let alone a mouse.  Ants, though, I think
>>we're starting to understand a little.
>>
>Is this kind of comparison really applicable to most systems? Ants can't
>do symbolic reasoning or learn to do imporve their checkers game. Perhaps
>in systems that attempt to do what ants do, it's reasonable, but otherwise,
>not so much. Just a thought.

     There's a giant gap in AI between the ant-level behavorial work
and the highly abstract systems based on predicates.  Attempts to fill
in this gap from the top down have been notably unsuccessful.  Attempts
to fill in this gap from the bottom up are progressing, but slowly.

					John Nagle
