Date: Wed, 16 Dec 92 14:57:12 EST
From: shapiro (Stuart C. Shapiro)
Subject: Definition of AI

Here is a revised version of the definition of AI for the
Undergraduate Brochure.  It is essentially what I use in the AI
article of the AI Encyclopedia:

Artificial Intelligence is concerned with the computational
understanding of what is commonly called intelligent behavior, and
with the creation of artifacts that exhibit such behavior.

Date: Wed, 16 Dec 92 15:39:37 EST
From: srihari (Sargur Srihari)
Subject: Re:  Definition of AI

This definition is as good as any.
Each leaves something to be desired.
Certain human activities are not associated with exhibiting intelligence
at all. Such as picking up a paper clip. Yet they are valid topics in AI.

Date: Wed, 16 Dec 92 19:32:23 EST
From: bharat (Bharat Jayaraman)
Subject: Re:  Definition of AI

Would such topics be considered as coming under Robotics?   Robotics is 
singled out as separate from AI these days, with phrases like 
"AI and Robotics".  At least Hopcroft and others seem to think so...

Date: Fri, 18 Dec 92 14:00:29 EST
From: shapiro (Stuart C. Shapiro)
Subject:  Definition of AI

I certainly believe in such a distinction.  It may well be that
picking up a paper clip is a part of Robotics and not a part of AI.
(Of course I have minimal control over what gets published in AI
journals and conferences.)

Date: Fri, 18 Dec 92 14:45:52 EST
From: srihari (Sargur Srihari)
Subject: Re:  Definition of AI

I was referring not to the motor aspects of the process of picking something
up but to the computational aspects. It involves vision, search and constraint
satisfaction (which are topics in anyones book on AI, including Stu's). 
More generally in AI we spend more time trying to get machines to do 
people-mundane things rather than people-smart things. 
If "intelligence" in the definition of AI
is used to refer to what living things (or dull people)do 
rather than what smart people do,
then the definition of AI is closer to reality. 
(A three-year old can recognize characters, which is what I work on,
 perhaps the NLP folks work on sentences understood by 4-year olds.
 At one time we used to work on chess-playing programs and expert systems,
 which is now passe)

Date: Fri, 18 Dec 92 15:13:28 EST
From: shapiro (Stuart C. Shapiro)
Subject:  Definition of AI

Ah.  I certainly agree that one of the things we learned in AI is that
some of the most impressive behavior of those people we regard as
particularly intelligent is easier to produce computationally then a
lot of the behavior that can be done by "any fool" (McCarthy's term).
Of course even Turing recognized that AI should be concerned with the
intelligent behavior of the average person (see his "superhuman human
fallicy" comment).  Nevertheless, for a short definition that is
supposed to communicate to freshmen, I think that is a good one (I
didn't make it up, but I forget where it came from---it's probably my
revision of something I heard from someone else.)

Date: Fri, 18 Dec 92 15:55:30 EST
From: srihari (Sargur Srihari)
Subject: Re:  Definition of AI

There seems to be agreement that Artificial Intelligence can also be
referred to as Artificial Stupidity.