Newsgroups: comp.ai.games
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!torn!watserv2.uwaterloo.ca!undergrad.math.uwaterloo.ca!sawhite2
From: sawhite2@undergrad.math.uwaterloo.ca (Steven White)
Subject: Re: AI routines for RPG game?
Message-ID: <D5zH35.Jx8@undergrad.math.uwaterloo.ca>
Sender: news@undergrad.math.uwaterloo.ca (news spool owner)
Nntp-Posting-Host: cayley.uwaterloo.ca
Organization: University of Waterloo
References: <D5E5oo.GDw@emr1.emr.ca> <3k3ddv$qc1@grivel.une.edu.au> <SMISHRA.95Mar14115808@kiwi.acns.nwu.edu> <3kfo5c$o6p@praline.no.neosoft.com>
Date: Sat, 25 Mar 1995 06:37:53 GMT
Lines: 30

In article <3kfo5c$o6p@praline.no.neosoft.com>,
Rick Abrams <ricka@praline.no.NeoSoft.com> wrote:
>
>I disagree. The problem with the above test is that 
>it was constrained. The definition of a Turing test
>I'm familiar with is to simply 'have a dialogue',
>not ask about subject 'X'. I think any system which
>can pass an unconstrained test *has to be* considered
>intelligent.
>
>-- 
>rha
On the other hand, a system that fails the test does not
necessarily fail the test of intelligence, or even simulating
a human.  After all, if I was the evil psychologist Jabba T. 
Hutt, I'm sure I could keep a kid in a room for a good many 
years, training him up, and while still quite intelligent, 
still likely would have trouble passing the Turing test.

And I'd suspect most AI's should they ever be made, would
likely not pass the test. Until you feel like getting your
synthetic people simulators. (Your own kid brother whom
you can put away in his box when you're tired of him. :) Most
of them would be somewhat alien. If only because they've not
had a child's background growing up.

So the Turin test is good for testing if you can fake a 
human. Not good for testing intelligence.

Steven White / After all, how much of Usenet's population fails the test? :)
