From newshub.ccs.yorku.ca!ists!torn!utcsri!rutgers!sun-barr!ames!ncar!noao!amethyst!organpipe.uug.arizona.edu!organpipe.uug.arizona.edu!bill Tue Jun 23 13:21:02 EDT 1992
Article 6294 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!torn!utcsri!rutgers!sun-barr!ames!ncar!noao!amethyst!organpipe.uug.arizona.edu!organpipe.uug.arizona.edu!bill
>From: bill@nsma.arizona.edu (Bill Skaggs)
Newsgroups: comp.ai.philosophy
Subject: Re: 5-step program to AI
Message-ID: <BILL.92Jun17163635@cortex.nsma.arizona.edu>
Date: 17 Jun 92 23:36:35 GMT
References: <1992Jun12.192537.32302@mp.cs.niu.edu> <60831@aurs01.UUCP>
	<1ljlk8d.nagle@netcom.com>
Sender: news@organpipe.uug.arizona.edu
Organization: ARL Division of Neural Systems, Memory and Aging, University of
	Arizona
Lines: 28
In-Reply-To: nagle@netcom.com's message of 17 Jun 92 06: 50:28 GMT

nagle@netcom.com (John Nagle) writes:

JN>	  Currently, we seem to be in the insect-level AI era.  Various 
JN>   researchers have demonstrated artificial insects of varying degrees of
JN>   functionality, and work is proceeding well.  Five years ago, nobody could
JN>   do a good insect; there's been real progress.  Brooks, Beer, and Travis
JN>   all have workable appraoches to engineering insects, and Maes and others
JN>   are now evolving insect-level nervous systems, if for rather dumb
JN>   insects.

I don't believe it makes sense to try to match AI to any particular
level of organism.  AI can currently build natural language
capabilities superior to those of a chimpanzee, but can't match the
motor control capabilities of a housefly.  So where could it fit?

This sort of comparison only makes sense if you restrict it to some
specific capability, such as visual pattern recognition, and even that
is considerably too broad.  A frog is probably as good at recognizing
bugs as any system that can currently be built, but it is hopelessly
bad at recognizing letters of the alphabet, even if they're
typewritten.

The work on artificial insects is extremely important, and it is
teaching us a great deal about how to do AI, but it does not mean we
are in the era of insect-level AI.  We are simultaneously below, at,
and above that level.

	-- Bill


