11.0028 Deep Blue, human intelligence and AI

Humanist Discussion Group (humanist@kcl.ac.uk)
Tue, 13 May 1997 19:34:05 +0100 (BST)

Humanist Discussion Group, Vol. 11, No. 28.
Centre for Computing in the Humanities, King's College London
<http://www.princeton.edu/~mccarty/humanist/>
<http://www.kcl.ac.uk/humanities/cch/humanist/>

[1] From: Mary Dee Harris <mdharris@acm.org> (16)
Subject: Re: 11.0022 Deep Blue, human intelligence and AI

[2] From: "Amsler, Robert" <Amsler@dyniet.com> (29)
Subject: RE: 11.0017 Deep Blue and human intelligence

[3] From: "Steven J. DeRose" <sjd@eps.inso.com> (25)
Subject: Re: 11.0022 Deep Blue, human intelligence and AI

--[1]------------------------------------------------------------------
Date: Tue, 13 May 1997 09:11:47 -0700
From: Mary Dee Harris <mdharris@acm.org>
Subject: Re: 11.0022 Deep Blue, human intelligence and AI

> "Does Deep Blue use artificial intelligence?
> The short answer is "no."I agree with Larry Taylor that this answer is inappropriate. Having
working in Natural Language Processing for many years and having to
defend it as a field of Artificial Intelligence, I learned to point out
that AI is not necessarily modelling human thought. It can be instead
developing a capability that requires human thought. Language is such an
example: no one knows how we humans are able to use language but we do.
Teaching a computer to use language then is artificial intelligence.

I don't know how to compare winning one set of games in computer chess to
a level of capability in a computer's language manipulation. Language
hasn't had the precise scorability of chess!

Mary Dee

-- 
Mary Dee Harris, Ph.D.			512-477-7213
Language Technology, Inc.		512-477-7351 (fax)
2415 Griswold Lane			mdharris@acm.org
Austin, TX 78703			mdharris@aol.com

--[2]------------------------------------------------------------------ Date: Tue, 13 May 1997 10:24:15 -0400 (EDT) From: "Amsler, Robert" <Amsler@dyniet.com> Subject: RE: 11.0017 Deep Blue and human intelligence

The results from Deep Blue beating Garry Kasparov are indeed a fitting subject for Humanist to discuss. In and of itself, one can dismiss this particular incident as atypical--another Grand Master may undoubtedly defeat THIS version of Deep Blue, but the handwriting is on the wall... The days when human beings can play better chess than a computer are drawing to a close. This is no more horrific than it was when steam engines were running in races with horses. Or rather, it is JUST as horrific; for that day ended the era of animal motor power being able to compete with machinery and claim victory. To be sure there are particular terrains which only an animal can carry a human--but earthmoving equipment is now the undisputed victor for work.

What Computers Can't Do... the debate unfortunately never really got off to a great start. The answer is simple. Computers can reason. They just can't think. Chess is one of the reasoning tasks humans have elected to elevate to higher significance than many others which are in fact far harder to perform; but it seems to be a matter of scale. Humans like chess because of its human dimensions. Great mathematicians used to spend time calculating the orbits of planets in their head; some even suggested "wasted" their time doing so given the likelihood that calculating devices would be invented that would perform the same task faster and more accurately; but now nobody would deem to devote their hours to trying to use their brain like a calculating machine.... We delegated the task to the computer. People play chess, however, because they like to, and therein lies the rub.

I suppose this is the fragile human ego. Because we LIKE to do something, we take offense when a machine takes the job over and does it better. Does this mean we have to stop playing chess? Well... we could invent a harder game if we really wanted to... but the issue is really just whether we can get used to the computer as a chess game judge and trainer and get back to doing what we do best---delegating reasoning tasks to the machines so we can create new ideas.

--[3]------------------------------------------------------------------ Date: Tue, 13 May 1997 12:06:14 -0400 (EDT) From: "Steven J. DeRose" <sjd@eps.inso.com> Subject: Re: 11.0022 Deep Blue, human intelligence and AI

At 10:14 PM 05/12/97 +0100, Larry Taylor wrote:

>> "Does Deep Blue use artificial intelligence? >> The short answer is "no." > >The short response is, "Ridiculous. Of course it's A.I. There >have been few problems more directly identified with A.I. since >the fifties than playing chess. Now that "we" -- ich bin ein >A.I. researcher! -- have WON, don't take it away from us!"

Agreed. But this is hardly new. Every time AI succeeds in some special domain, that domain becomes "no longer AI". C'est la vie.

>Repeat after me: Since May 11, 1997, a computer is the best chess >player in the universe. It can't talk, think about anything >other than chess, or tie its shoes, but it is better at what it was >engineered to do than ANY HUMAN ALIVE.

Again, true. But hardly significant by itself. Many a simpler machine has been "better at what it was engineered to do than ANY HUMAN ALIVE". For example, your everyday electric drill. Or a pencil. Deep Blue is a superb accomplishment, but a comparison to human intelligence is like comparing apples to poetry (one might entertain a comparison to an idiot savant). Optimizing for any single purpose is relatively easy.

>Particularly significant was Kasparov's whining after his loss, >which showed that his ego has been crushed.

If Deep Blue had a sound chip, would it have whined last year?

Steven J. DeRose Chief Scientist Inso EPS (formerly EBT)