Newsgroups: comp.ai.games
Path: cantaloupe.srv.cs.cmu.edu!nntp.club.cc.cmu.edu!godot.cc.duq.edu!newsfeed.pitt.edu!gatech!news.mathworks.com!news.kei.com!wang!news
From: bruck@actcom.co.il (Uri Bruck)
Subject: Question of AI (was : DUMB QUESTION?)
Reply-To: bruck@actcom.co.il
Organization: ACTCOM - Internet Services in Israel
Date: Fri, 1 Dec 1995 09:39:51 GMT
Message-ID: <DIwIzM.11A@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> <DIJKyq.AKL@actcom.co.il> <49e6pi$4q2@news.worldlinx.com>
Sender: news@wang.com
Lines: 87

sshah@intranet.on.ca wrote:

>Reply-to: Bruck@actcom.co.il
>Reply-from: Unknown
>Subj: Re: DUMB QUESTION?

>Hi!

> Br> The original Turing test actually has three seperate rooms, one for
>[...]
> Br> about the nature of the interrogated. This is done to avoid prejudice
>[...]
> Br> to do with thinking, and suggested that intelligence should be judged
> Br> on its own merits.

>Personally, I'm no proponent of the Turing test.  I agree with the premise,
>but the implementation is faulty.  Who says that human speech has anything to
>do with intellience?  Thousands of languages exist, why not have the computer
>have its own?

Nobody says that - that is the other side Turing's coin, at least from
I'm looking, Turing says that we can formulate an opinion on something
we can understand and have a common language, whether it be a human
language, or specific jargon, with.
I've seen someone put forward the claim, might have even been in this,
or in a related forum, that grand-master class chess programs don't
pass a Turing test analog limited to the chess-game domain. The
argument was based on grand-master's claims that computer-chess
programs have a style different than any human player, thus making
them distinguishable from humans.

>Anyway I'm not a big supporter of the AI push either.  I'm a practical sort of
>guy who sees AI research as interesting but then immediately tries to apply it
>to something practicle.
In that sense I see AI research as a success.

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

>Yes.  That's another problem.  The Turing test is more concerned with
>deceiving.  I'm not sure where this is going.  Perhaps I'd prefer a test that
>would quantatively measure the abilities of the AI instead of watching it
>qualitatively fool a person into thinking it was intelligent.

I don't think it's concerned mainly with deceiving, it's concerned
with  giving the artificial inteligence the same chance that a human
has to be socialy accepted, without prejudice because of different
shape, origins, or lifestyle. 

When you meet people, and talk to them, you don't quatitavely measure
their intelectual abilities before you accept them as partners for
conversation. 

> Br> competition demands any program to be able to converse inmore than one
> Br> specific area, while Turing required the machine to be able to reply

>The problem is that most people CAN'T talk in every area.  Sometimes we just
>stand there with our tongues to the ground speechless when confronted with a
>topic we don't have a clue 
You cut out the relevant Turing qoute which deals with that
"
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.
"
Perhaps we can't always give answers as good as that, but Turing
expected an AI to forego the need for great big knowledge-bases with
the very human trait of avoiding the question, or simply admitting
ignorance for that subject


> >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? :)
> Br>                                                    ^^^^^^^^^^

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

>Not many. :-)  Hopefully you got the joke instead of thinking that I was the
>one who was a big joke. :)
You need not worry, I did not think you were the joke, you have passed
my t.Uri.ng test long ago ;)

> Br> Who is qualified to be an interrogator???

>Another AI.

I smell a reductio ad absurdium

