COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	aaai[f82,jmc]		To do as president
C00006 ENDMK
C⊗;
aaai[f82,jmc]		To do as president

1. A quarterly "president's message".

a. Let's work on the fundamental problems of AI - not so much on tools
for debugging and on demos.  We need to discuss what we think of the
progress and not spend our time arguing with outsiders.
The sponsors are distorting the field, but it's really more
our fault than theirs.  Not all problems are
equally suited for scientific progress.


I'll tell you what I think the fundamental problems are, but if you
differ, I urge you to decide what you think is fundamental and work
on that.  Chess is the drosophila?

b. The orderly expansion of the field.  No quick tenure.  Keep
room for newcomers.

c. What should AAAI be?  (1) A scientific organization and not
a trade union, i.e. rewards the best work rather than advances
everyone in the field.

d. Presidential address.  Epistemological problems of AI.
The search for generality in AI.

e. What does a university need for ai research?  What should be
taught?

f. Criteria for testing and comparing theories in AI.  Chess as
sport and as a drosophila.  Compare VanLehn, Kurt, John Seely Brown
and James Greeno
"Competitive Argumentation in Computational Theories
of Cognition", to appear in Methods and Tactics in Cognitive Science,
W. Kinsch, J. Miller and P. Polson (Eds.) New York: Erlbaum

Perhaps I should try to meet various people:
nsf, arpa, afosr,onr, aro, ai in medicine, acm, ieee pattern, ams, aaas
nat. acad., keyworth

Perhaps there should be a small journal, AI reviews, that would critically
review important papers and would help form public opinion on what
constitutes good work in ai.
What consitutes verification of ai ideas? van Lehn at xerox.

conferences:
epistemology, common sense, planning (Sacerdoti), curriculum, standards
for judging work

prizes: not me

Rick Hayes-Roth and Bob Englemore want a more "professional" organization.
Their clients found AAAI-82 unhelpful and the presentations
not very good.

1983 July 3
The most commonly misused word in computer science is "understand".  A
good explanation of the varieties of understanding will be important.