COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002		It  seems to me that the Artificial Intelligence and
C00006 ENDMK
C⊗;
	It  seems to me that the Artificial Intelligence and
Philosophy group was, as a group, only a partial success.
I had hoped that the interaction would lead to a sharp formulation
of epistemological problems that artificial intelligence work
would have to solve in order to make programs with intelligence
qualitatively similar to that of humans.  It seems to me that this
goal was not achieved.

	Perhaps the goal was unrealistic in that too much preliminary
work has to be done before philosophers and AIers cans speak enough
of the same language, but it didn't seem to me that the communication
difficulties were primarily across field boundaries.  Previous commitments
to finish off certain research topics seemed to play a more important
role.  Since many of my own previous research commitments were precisely
in this area, I found myself somewhat impatient.

	My own work in this area during the year involved the following:

	1. I finished my paper "Circumscription - A Form of Non-Monotonic
Reasoning" which is being published in the journal Artificial Intelligence.

	2. I continued work in non-monotonic reasoning beyond the paper,
clarifying the relations between circumscription and the non-monotonic
"logics" of McDermott-Doyle and Reiter.

	3. I formulated the idea that non-monotonic reasoning is
used  by humans and must be used by machines to avoid making all possible
philosophical distinctions.   This notion is akin to the notion that
the treatment of meaning in philosophy cannot depend on the previous
solution of all the problems of science.

	4. I now see a way to use first order logic to treat concepts
as objects and still avoid a notational distinction between opaque
and non-opaque occurrences of terms.  The goal is that a computer
program should be able to treat knowing a telephone number and
dialing a telephone number in entirely parallel ways until the
distinction becomes important.

	Towards the end of the year, perhaps as the previous
commitments were fulfilled or abandoned, it seems to me that the
group discussions became more effective and began to contribute
to the solution of the common problems of artificial intelligence
and philosophy.