What is AI?

Dan Schwartz <schwartz@iota.cs.fsu.edu>
Date: Wed, 15 Sep 93 20:40:10 -0400
From: Dan Schwartz <schwartz@iota.cs.fsu.edu>
Message-id: <9309160040.AA01022@iota.cs.fsu.edu>
To: interlingua@ISI.EDU
Subject: What is AI?

Have been thinking some more about (i) my proposed definition of AI,
(ii) John McCarthy's criticism of it, and (iii) my attempted
clarification---and have decided I need to retract part of my last
argument.

The definition proposed was:  AI is that field of endeavor which is
concerned with understanding the activities of the human mind and
simulating those activities on a computer.

And my clarification amounted to elaborating on my intended meaning of
"simulate" and drawing up an analogy of bull dozers replacing humans for
purposes of moving dirt.

I'm now convinced that "simulate" is just the wrong word here.  The
reason is that to simulate something means to mimic that thing in the
form of an abstraction, whereas bull dozers actually do move (real)
dirt, and computerized reasoning systems actually do perform (real)
if-then reasoning.

Thus a modified definition I'm now considering would be the above with
"simulating those activities on a computer" replaced by "performing
those same activities on a computer," with the understanding that the
underlying mechanisms (brain vs computer) are widely different.

In making this change, however, I find myself once again asking what is
indeed "artificial" about Artificial Intelligence.  Surely, if computers
are made to do _real_ if-then reasoning, then this doesn't warrant being
referred to as "artificial".  This also becomes an incorrect choice of
terms.

In turn this leads to the question as to whether if-then reasoning
deserves to be called "intelligence," or for that matter, whether what
we are dealing with in AI resembles true intelligence at all.  My belief
is that it does, albeit in a very limited form.  Further, I claim that
one of the aims of AI is to make our computer systems increasingly more
intelligent, by getting them to do even more of the types of things we
do with our brains (as well as some things we would like do with our
brains, but are not actually able to).

In recent years, many have suggested that the term "Artificial
Intelligence" is outmoded, and we should come up with a new term for
this domain.  I've often thought "Automated Reasoning" would be better,
but this was appropriated early on by the automated theorem provers, and
anyway, there does seem to be more to consider than just logical
reasoning.  For example, it may not be long before we have machines that
can detect and respond appropriately to human emotions, or for that
matter, exhibit emotion-like behavior.  Shouldn't this research also
fall under "AI"?

How about letting "AI" stand for "Automated Intelligence"?

Comments anyone?

--Dan Schwartz