Xref: utzoo comp.ai.philosophy:1066 comp.ai:9557
Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!cis.ohio-state.edu!ucbvax!dev.sas.com!sasmjw
From: sasmjw@dev.sas.com (Martyn Wheeler)
Newsgroups: comp.ai.philosophy,comp.ai
Subject: Re: Turing Test, what's the point?
Keywords: AI, computers
Message-ID: <9106261905.AA02183@hawkwind.dev.sas.com>
Date: 26 Jun 91 19:05:07 GMT
References: <612@ckgp.UUCP>
Sender: daemon@ucbvax.BERKELEY.EDU
Reply-To: sasmjw@dev.sas.com
Organization: SAS Institute Inc. Cary NC
Lines: 38

The formulation of the Turing Test had little to do with Turing's
ability to predict computer technology.  After all, all the test
really does is to give an arbitrary subjective measure of the ability
of one system to imitate another.
   If the observer cannot determine which is human and which is
computer, who is to say that it is not the human who is successfully
simulating the computer rather than the other way round?  Is it
perhaps even a test not of the computer and human subjects, but rather
of the discriminatory abilities of the observer?  What if the observer
is in fact a computer trying to distinguish between two human
subjects?
   The point of the Turing Test is surely that it provides a fixed
starting point that could be used to anchor the field of computer
intelligence when the science was (is?) young.  Its value is not so
much that it provides a test of intelligence itself, but that it leads
to so many interesting questions.
   An actual Turing Test would be much more valuable to the
meta-observer, who can examine the actual observer's behaviour as well
as that of the two subjects, than to anyone directly involved in the
experiment itself.
   The ability of a computer to "pass" the Turing Test is little more
than a measure of a computer's ability to pass the Turing Test, and
extending that to an assumption of intelligence is an unjustified leap
of faith.  Achieving a simulation of something is not the same as
achieving the thing itself, and we deceive ourselves if we believe
that by working towards a simulation we are necessarily making any
progress towards the real thing.
   There are times in the real world, however, when a simulation is
all that is required.  In the fields of commercial Natural Language
and Machine Translation, for example, it is not necessary to believe
that one is duplicating a human, nor even should this be a goal -- an
automated Chinese Room is quite sufficient for useful applications.

Martyn
------------------------------------------------------------------------
sasmjw@dev.sas.com      "If you spin, you deserve to die"--Mike Hawthorn
(919) 677-8000 ext.7954 (SAS Institute, Inc., Cary, NC)
(919) 839-0092 (h) (Raleigh, NC)              Martyn Wheeler -- DoD #293