Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site ames-lm.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!mhuxl!houxm!houxz!vax135!cornell!uw-beaver!tektronix!hplabs!ames-lm!eugene
From: eugene@ames-lm.UUCP (Eugene Miya)
Newsgroups: net.ai
Subject: Re: Philosophy and other amusements.
Message-ID: <383@ames-lm.UUCP>
Date: Thu, 5-Jul-84 13:10:42 EDT
Article-I.D.: ames-lm.383
Posted: Thu Jul  5 13:10:42 1984
Date-Received: Sat, 7-Jul-84 01:37:30 EDT
References: <975@eosp1.UUCP>
Organization: NASA-Ames Research Center, Mtn. View, CA
Lines: 21

With regard to expert systems, I thought of an interesting
[take this with a grain of salt] set of tests to evolve or refine
the development of such systems.  These tests would test the expertise
of such systems.  Take a classic system like MYCIN.
When the developers feel the system is ready for a shake down,
[remember, this is not entirely serious, but not for the weak of heart]
infect the developers of the system with one of the diseases in the
knowledgebase, and let them diagnose their own ailment.
There might be interesting evolutionary consequences in software development.

Similarly, other people developing other systems would put their
faith and lives on the line for the software systems they develop.
Are these systems, truly 'expert?'

Admittedly, not a rigorous test, but neither was Turing's.

The above are opinions of the author and not the funding Agency.

--eugene miya
  NASA Ames Research Center
  {hplabs,hao,dual}!ames-lm!aurora!eugene