COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	Prof. Paul C. Martin
C00010 ENDMK
C⊗;
Prof. Paul C. Martin
Division of Applied Sciences
Harvard University
217 Pierce Hall
Cambridge, MA 02138

Dear Professor Martin:

	I am familiar with all the people you mention, but I don't closely
follow the papers of any of them.  All are better known than the people
you mention as being on the Harvard faculty with interests in artificial
intelligence.

	I suppose Drew McDermott has the highest scientific potential
of the group.  He has worked on quite a variety of hard problems and
has gotten significant results.  He also exercises a reasonable degree
of self-criticism, so he knows when one of his attempts is not
successful enough.  Roger Schank has ideas and energy, but often
pushes inadequate ideas beyond their merits.  He supervises large
numbers of students.  I would worry some about his commercial interests.
Mike Brady has a good reputation in robotics, but I have no first hand
familiarity with his work, having lost interest in the field.  I am
not familiar with Randall Davis's recent work which I suppose is in
expert systems.  As far as I know Feldman's recent activity has been
mainly administrative.  When he was at Stanford, I did not consider
him extremely creative.

	In my opinion you should look beyond this group - at least
I suppose Harvard can make offers that are interesting to the best
people in any field.  Robert Boyer and Jay Moore (a team) have done
the best work in the world in automatic and interactive theorem proving.
Tom Mitchell at Rutgers and Ryszard(?) Michalski (separately) have done
the most interesting recent work in machine learning.  Robert Kowalski
of Imperial College and Alain Colmerauer of the University of
Aix-Marseille are the originators of logic programming.  Strong
younger people include Fernando Pereira and Stanley Rosenschein
of SRI International.  If you could get Gerald Sussman of M.I.T.,
that would be a great catch.  We recently got Nils Nilsson of SRI
to be our department chairman.  Among older people, Donald Michie
has great initiative and would get many good things started.  He
got into some fights through pushing AI harder than the British
system would tolerate, but I think he could be a big help.

	I summary the people you mention are ok, but you should
cast your net wider.  Perhaps you need more high level advice in person
from people who know who's who.  Nilsson,
Michie, Sussman and/or Boyer would be good for that.

Sincerely,