COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	questi[e84,jmc]		AI questionnaire from Bobrow and Hayes
C00006 ENDMK
C⊗;
questi[e84,jmc]		AI questionnaire from Bobrow and Hayes

I suppose I'd like one more crack at these answers before
you actually publish anything.  Perhaps you should think again
about whether a 25th anniversary celebration is the best use
of the pages.

1. The main scientific advance that I can identify is the
formalization of non-monotonic reasoning.  Naturally this
may be a prejudiced view.  Perhaps the progress in learning
amounts to a major advance.  I agree with Simon that more
progress would have been made if more people took its advantages
as a scientific problem seriously.  The current chess programmers
seem to treat it more as a sport and don't really study the
conceptual weaknesses of their work.

2. Circumscription is my own best work.  All the major problems
of AI seem untractable.

3. I see no limits, but I can't predict where there will be the
most progress, because I see the requirements as conceptual.

4. Both.  Many will be disappointed, but there may be some solid
applications of AI that will revolutionize some fields.

5. I think there will be a unified AI methodology, but we don't
have it yet.  Of course, the conceptual split into basic research
and applications is relevant to AI as well as to other fields.
Failure to make this split clearer has led to  weakness in basic
research.  For this research sponsors bear some of the responsibility.
See my AI Magazine President's message on the subject.

6. I don't see any problems that will become easy just because of
more processing power.  Given the current ideas, we have just
recently (at Stanford anyway), got computing resources adequate
for the current projects.

7. I don't think the serial/parallel distinction will be of much
importance in the theory of AI.

8. It is best to postpone consideration of such issues until they
actually arise.  Current social thought is much to inclined
to shy at paper bags for there to be any likelihood that any good
will come of rounding up the usual interdisciplinary suspects
once again.

9. AI will absolutely revolutionize human society sometime in
the next 5 to 500 years.  However, only actual developments will
attract attention of the right sort.  Imagine asking the Reagan
and Mondale campaigns for policy statements on AI.  What garbage
would be produced by the third assistant speech writer!

10. I don't know any questions that shouldn't be asked.  The right
questions are about what are the basic intellectual mechanisms.