Newsgroups: comp.ai.games
Path: cantaloupe.srv.cs.cmu.edu!bb3.andrew.cmu.edu!nntp.sei.cmu.edu!cis.ohio-state.edu!math.ohio-state.edu!uwm.edu!lll-winken.llnl.gov!simtel!news.kei.com!wang!news
From: bruck@actcom.co.il (Uri Bruck)
Subject: Re: DUMB QUESTION?
Reply-To: bruck@actcom.co.il
Organization: ACTCOM - Internet Services in Israel
Date: Fri, 24 Nov 1995 09:53:45 GMT
Message-ID: <DIJKyq.AKL@actcom.co.il>
References: <DH1BCp.180@ricks.edu> <47cudn$1of@bs33n.staffs.ac.uk> <DHu3EI.5G5@actcom.co.il> <483av4$6dt@mozo.cc.purdue.edu> <48j10j$nr3@news.worldlinx.com>
Sender: news@wang.com
Lines: 56

sshah@intranet.on.ca wrote:


>The Turing Test isn't all that specific.  It just said that if a human could
>have a conversation with a computer and it sounded real enough it passed.  He
>didn't say anything about terminals or tricking the human into thinking it was
>another human from the get go.

The original Turing test actually has three seperate rooms, one for
the interrogator, one for another person, and the third one for the
mcahine. The interrogator is supposed to communicate with them, in
such a way that the communication medium will provide any information
about the nature of the interrogated. This is done to avoid prejudice
on the part of the interrogator, since Turing supposed that creating
something that looks like a human, and sounds like a human has nothing
to do with thinking, and suggested that intelligence should be judged
on its own merits. 

"...The form in which have set this problem reflects this fact in a
condition which prevents the interrogator from seeing or touching the
other competitors or hearing their voices."

The goal of the machine is to indistinguishable from a human.
"In any case there is no intention to investigate here the theory of
the ganme, and it will be assumed that the best strategy is to try to
provide answers that would naturally be given by a man."

So tricking the interrogator is a part of the Turing test as presented
inthese qoutes from Turing's "Computing machinery and Intelligence"



>There are Fake Turing Test Contests all around.  Don't be fooled, they tend to
>be completely worthless.

Because it is usually the human who manages to sound like a machine
more than the machine manages to sound like a human. anyway, no
competition demands any program to be able to converse inmore than one
specific area, while Turing required the machine to be able to reply
to questions, or at least take verbal evasive action, on any subject.

"
Q:Please write me a sonnet on the subject of the Forth Bridge.
A:Count me out on this one. I never could write poetry.
"

>Maybe having a conversationalist computer isn't such a good idea anyway.  Why
>should it speak OUR language?  Why not us its? <-- Is that sentence? :)
                                                      ^^^^^^^^^^

How many of us can pass the Turing test    BIG ;)       ????

Who is qualified to be an interrogator???

Uri

