From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!usc!elroy.jpl.nasa.gov!ames!olivea!uunet!mcsun!uknet!edcastle!aisb!jeff Tue Jan 28 12:17:32 EST 1992
Article 3131 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!usc!elroy.jpl.nasa.gov!ames!olivea!uunet!mcsun!uknet!edcastle!aisb!jeff
>From: jeff@aiai.ed.ac.uk (Jeff Dalton)
Newsgroups: comp.ai.philosophy
Subject: Re: Table-lookup Chinese speaker
Message-ID: <1992Jan24.193202.9713@aisb.ed.ac.uk>
Date: 24 Jan 92 19:32:02 GMT
References: <1992Jan23.160551.21216@oracorp.com> <1992Jan23.231248.40983@spss.com>
Sender: news@aisb.ed.ac.uk (Network News Administrator)
Organization: AIAI, University of Edinburgh, Scotland
Lines: 46

In article <1992Jan23.231248.40983@spss.com> markrose@spss.com (Mark Rosenfelder) writes:
>The business of confining the human or machine to be tested to a small 
>room seems to stack the deck even further in favor of the machine.

Look, do you want a test for understanding, or a test for whether
someone is in a room that also contains a clock and some recent
newspapers?

>>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.
>
>The program with the time stamp is not the simple lookup table any more.

Ok, so it's the simple-table-lookup-with-clock example instead.

>But besides this, there are other questions we can ask to stump the
>table-lookup machine.  "What city are we in?"  "What year is it?"

Why would these stump it?  There are clocks that tell the year,
by the way.  And we have to assume the system can answer questions
like "how old are you?" and "where were your parents born", so
why not "what city?".

>The table can contain as sensible a set of replies to these questions as
>we can find, but an accumulation of such evasive answers is going to
>be very suspicious.

They don't have to be evasive.  Remember it's not a test for whether
the program tells the truth.

>Conversing with the table-lookup machine would be kind of like talking
>with a person suffering from advanced Korsakov's syndrome-- someone who
>has no ability to remember events past a certain point in his life.
>I wouldn't equate being able to simulate a mental patient with "passing
>the Turing test."  Is PARRY intelligent?

But such things may well be problems for programs in general.
If we restart them, they don't remember what they did before,
and so on.  As for things that came up a while back in the
conversation, by supposition the table lookup machine has that
stuff in the table.

-- jd


