From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rpi!think.com!spool.mu.edu!uunet!psinntp!scylla!daryl Thu Jan 16 17:20:11 EST 1992
Article 2690 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rpi!think.com!spool.mu.edu!uunet!psinntp!scylla!daryl
>From: daryl@oracorp.com
Newsgroups: comp.ai.philosophy
Subject: Re: Intelligence testing
Message-ID: <1992Jan14.045132.19208@oracorp.com>
Date: 14 Jan 92 04:51:32 GMT
Article-I.D.: oracorp.1992Jan14.045132.19208
Organization: ORA Corporation
Lines: 27

Mark Rosenfelder writes:

>>On the other hand, if you ignore such practical arguments, and assume
>>that computers can be made arbitrarily fast, and have arbitrarily much
>>memory, then it follows immediately that a computer could pass the
>>test for being able to converse in Chinese with the fluency of a
>>native. There are only a finite number of possible sensible
>>conversations in Chinese in the lifetime of a human being. The
>>computer could store all of these, and do a simple table look-up, as
>>someone (perhaps you) has pointed out in the past.

>There may well be non-intelligent ways to pass the Turing Test, but this
>isn't one of them.  One could not look up an entire conversation _while
>it is still going on_, no matter how fast the computer.  At any point in
>the conversation there is a huge number of possible conversations which
>duplicate it up to that point and diverge thereafter.

The fact that there are a huge number of conversations consistent with
what has been said so far is irrelevant: the computer only needs to
select a response that occurs in *some* conversation that agrees with
the current conversation so far. If more than one response is possible,
the computer just picks one. Conversations don't have to have a unique
outcome.

Daryl McCullough
ORA Corp.
Ithaca, NY


