From newshub.ccs.yorku.ca!torn!utcsri!rpi!usc!wupost!uunet!sequent!muncher.sequent.com!bfish Thu Jul  9 16:20:38 EDT 1992
Article 6431 of comp.ai.philosophy:
Newsgroups: comp.ai.philosophy
Path: newshub.ccs.yorku.ca!torn!utcsri!rpi!usc!wupost!uunet!sequent!muncher.sequent.com!bfish
>From: bfish@sequent.com (Brett Fishburne)
Subject: Re: Defining other intelligence out of existence
Message-ID: <1992Jul8.170459.23090@sequent.com>
Followup-To: comp.ai.philosophy
Sender: bfish@sequent.com
Nntp-Posting-Host: sequent.sequent.com
Organization: Sequent Computer Systems Inc.
References: <4eKPF7_00WAL01Wvsr@andrew.cmu.edu> <1992Jul7.214542.13861@oracle.pnl.gov>
Date: Wed, 8 Jul 92 17:04:59 GMT
Lines: 57

In article <1992Jul7.214542.13861@oracle.pnl.gov> d3g637@duke.oname writes:
>In article 00WAL01Wvsr@andrew.cmu.edu, bh1q+@andrew.cmu.edu (Bob Troyer Hansen) writes:
	<beginning of post deleted>
>
>Quite right.  What I was getting at is more along the lines of:
>
>	X passes the test -> X communicates intelligently
>
>The reason for my concern is that it seems quite a bit easier, in my
>opinion, to write a program that converses intelligently instead
>of being intelligent.
>

BRAVO! I wholeheartedly agree! A small modification, however,

	X passes the test -> X communicates like an intelligent *human*

In a previous post I put forth the argument:
>>	Intelligence is distinct from knowledge.

David Chassin argues:
>
>which leads me to think Turings test might be better stated:
>
>	X passes the test AND X uses knowledge -> X is intelligent
>
>Any thoughts?

I disagree.  I think that the previous description accurately defines the
Turing Test.  I am willing to argue that:

	X is intelligent AND X uses knowledge -> X passes the test

or more importantly:

	X is intelligent AND X uses knowledge -> X comunicates like an
		intelligent *human*

The important point being that the KNOWLEDGE used is what it is like to be
human.  Clearly, this is the approach computer programmers use in attempting
to pass the Turing Test.  A program is given a database of human knowledge and
then (using appropriate algorithms) assembles that knowledge into communication
patterns which are human.  The INTELLIGENCE involved is trapped in the 
algorithms, not in the silicone.  As a result, I would argue that attempting
to generate an artificially intelligent machine by attempting to meet the
conditions of the Turing Test will inevitably FAIL.  The Turing Test can be
passed by an intelligent machine (with the appropriate knowledge), but here
the emphasis should be in generating an intelligent machine and gathering the
information necessary for the database.  Note that a prerequisite for
success is _generating an INTELLIGENT machine_.  To some extent, the Turing
Test is verfication of intelligence -- but only if the proper knowledge has
been gathered.  Thus, a truly intelligent machine could fail the Turing Test
because of inadequate knowledge, which implies that the Turing Test is _not_
a good assessment for intelligence.

-- Brett



