Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.3 4.3bsd-beta 6/6/85; site uw-june
Path: utzoo!watmath!clyde!burl!ulysses!bellcore!decvax!tektronix!uw-beaver!uw-june!jon
From: jon@uw-june (Jon Jacky)
Newsgroups: net.ai
Subject: Re: Technology Review article
Message-ID: <429@uw-june>
Date: Wed, 19-Feb-86 23:27:50 EST
Article-I.D.: uw-june.429
Posted: Wed Feb 19 23:27:50 1986
Date-Received: Fri, 21-Feb-86 07:39:57 EST
References: <3189@umcp-cs.UUCP>
Distribution: net
Organization: U of Washington Computer Science
Lines: 22
Summary: "After 25 years AI has failed to live up to its promise" is correct.

> (Technology Review cover says...)
> After 25 years Artificial Intelligence has failed to live up to its promise
> and there is no evidence that it ever will.

Most of the comment in this newsgroup has addressed the second clause in 
this provocative statement.  I think the first clause is more important, and
it is indisputable.  The value of the Dreyfuss brothers' article is to
remind readers that when AI advocates make specific predictions, they are 
often over-optimistic.  Personally, I do not find all of the Dreyfuss'
speculations convincing.  So what?  AI work does not get funded 
to settle philosophical arguments, but because the funders hope to derive
specific benefits.  In particular, the DARPA Strategic Computing Program,
the largest source of funds for AI work in the country,
asserts that specific technologies (rule based expert systems, parallel
processing) will deliver specific results (unmanned vehicles that can
drive at 40 km/hr through battlefields, natural language systems with 
10,000 word vocabularies) at a specific time (the early 1990's).  One
lesson of the article is that people should regard such claims 
skeptically.

Jonathan Jacky,  	...!ssc-vax!uw-beaver!uw-june!jon  or jon@uw-june
University of Washington