Xref: utzoo comp.ai:2602 talk.philosophy.misc:1566
Path: utzoo!utgpu!watmath!clyde!att!whuts!homxb!homxc!marty
From: marty@homxc.UUCP (M.B.BRILLIANT)
Newsgroups: comp.ai,talk.philosophy.misc
Subject: Re: Artificial Intelligence and Intelligence
Summary: unsupported allegations
Message-ID: <4216@homxc.UUCP>
Date: 15 Nov 88 15:55:53 GMT
References: <484@soleil.UUCP>
Organization: AT&T Bell Laboratories, Holmdel
Lines: 66

In article <484@soleil.UUCP>, peru@soleil.UUCP (Dave Peru) writes:
> Definition of Intelligence:
> 
> 1. Know how to solve problems.
> 2. Know which problems are unsolvable.
> 3. Know #1, #2, and #3 defines intelligence.

Definitions are arbitrary.  My only criteria for definitions are
whether they create useful terms, and whether they conflict with
previously accepted definitions.  This one might be useful, conflicts
with other definitions, but I'm willing to try it out.

One problem with it is the use of the word ``know.''  There is no such
think as ``knowing'' a definition abstractly.  Definitions change, as
people change the way they use words, because they change the way they
use the ideas the words represent.  A definition that is useful one day
may be found later to be inadequate or self-contradictory.

> This is the correct definition of intelligence.  If anyone disagrees, please
> state so and why.

I disagree.  There is no such thing as a correct definition, only
useful or useless.  The definition is interesting, at least, but, as
noted above, it has problems.

There is already a vague concept of ``intelligence'' waiting for a
precise definition.  The concept contains the notion that human
behavior is characterized by some property called ``intelligence''
which is absent from the behavior of most other living things and most
machines.  That notion prejudices any attempt to define artificial
intelligence, because it presupposes that machines are not intelligent.

Any definition of ``artificial intelligence'' must allow intelligence
to be characteristically human, but not exclusively so.

> If you take into account the unsolvable problems of Turing machines, then this
> proves Artificial Intelligence is impossible.
> 
> "Artificial Intelligence" is an unsolvable problem.

This is a statement of a proposition and a preface to a proof.  It is
not a proof.  Proof required.

> Human beings are not machines.

Therefore..... what?  Is that supposed to mean that all machines are
bound by certain unsolvability rules derived from the study of Turing
machines, but the human mind is exempt?  Can that be proved?

> Human beings are capable of knowing which problems are unsolvable, while
> machines are not.

Proof?  It seems to me that all we humans have is an existence proof that
some problems are unsolvable.  Classifying all problems into solvable and
unsolvable may be itself an unsolvable problem - does anybody know?

We are, I think, back to the primitive notion that (a) intelligence is
what only humans do, (b) machines are not human, hence (c) machines are
not intelligent.

M. B. Brilliant					Marty
AT&T-BL HO 3D-520	(201)-949-1858
Holmdel, NJ 07733	att!houdi!marty1

Disclaimer: Opinions stated herein are mine unless and until my employer
	    explicitly claims them; then I lose all rights to them.