From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rutgers!usc!wupost!uunet!psinntp!scylla!daryl Tue Jan 28 12:16:33 EST 1992
Article 3059 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rutgers!usc!wupost!uunet!psinntp!scylla!daryl
>From: daryl@oracorp.com
Newsgroups: comp.ai.philosophy
Subject: Re: Table-lookup Chinese speaker
Message-ID: <1992Jan23.160551.21216@oracorp.com>
Date: 23 Jan 92 16:05:51 GMT
Organization: ORA Corporation
Lines: 68

Drew McDermott says:

> I can't pass up a chance to agree with Mikhail Zeleny

Its nice to occasionally be able to say something that both the
pro- and anti-AI side disagrees with.

> All we have to do to defeat the table is allude to some fact that is
> true now and known to the average human, but not always true.  E.g.,
> the fact that earth is inhabitable.  Or, better yet, pick a fact that
> might be true at some times but is not true now.  Suppose we begin our
> dialogue thus:
> 
>    "Isn't it sad how all life on earth was wiped out a few years ago?"
> 
> and continue with other absurdities.  The lookup system will give
> itself away by responding with things like:
> 
>    "Perhaps we can find a way to cause life to re-evolve."
>     or
>    "Oh, well, life was overrated, in my opinion."
>     or
>    "Tell me more about life on earth."
> 
> whereas a normal person would say,
> 
>    "Are you out of your mind?"

I believe that you misunderstood the premises. The table's responses
only include what a normal person would respond. You are certainly
correct that this is time-dependent, but by that criterion, William
Shakespeare wouldn't pass the Turing Test, either, since he would have
no sensible response to questions about airplanes or atomic power.

> Perhaps the thought experiment is supposed to involve having the table
> get updated every few seconds to reflect changes in the world.  If so,
> one must begin to inquire about the agency that is doing the updating,
> and what its (rather stressful) mental life might be like.

I thought that the updating mechanism was quite clear. The Chinese
Room is supposed to simulate a person confined to a small room, with
no way of communicating with the outside word except through a textual
interface. The only "updating" that is possible is by telling the man
in the Chinese Room new facts. So for a Chinese Room that is supposed
to model a present-day human, the only Turing-Test passing response
to "Do you think the king of the United States will succeed in having
his daughter succeed him?" is "There is no king of the United States!"
(or something like that). After Bush is reelected or defeated, the CR
will not know that unless someone tells it, which is the same as for
a human confined to a single room.

You are right that the question of "what time is it" would stump the
Chinese Room, since no sense of time was built in. However, that is
addressable by either 1. Having the CR say "I don't have a watch, and
my sense of time in here is unreliable", or 2. Having each input
message be accompanied with a time-stamp telling the current time.

If this latter solution violates the spirit of the original
suggestion, then I have to admit defeat, since obviously humans have a
sense of time passing. The point of the Turing Test (and the Chinese
Room) was supposed to be to create a situation where senses played no
role, only verbal information. That was the issue that I was addressing,
whether the verbal reasoning ability of humans could be captured by
a table-lookup. Obviously sensory abilities cannot be so captured.

Daryl McCullough
ORA Corp.
Ithaca, NY


