COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	.require "let.pub[let,jmc]" source
C00009 ENDMK
C⊗;
.require "let.pub[let,jmc]" source;
∂AIL Dr. Robert Marston, President↓University of Florida
↓Gainesville, Florida∞

Dear Dr. Marston:

	Many thanks for your hospitality on your island and at your
home and for so patiently letting Marvin and me harangue you.
I enclose the enclyclopedia article on AI that I promised and
also, in case you may find it interesting, an article ≥hat covers
some of the common area between artificial intelligence and
philosophy.  I send it, because it elaborates my reasons for believing
that artificial intelligence and philosophy have more in common
than the usual connection between philosophy and a science, the
philosophy of which may be studied.

	Here are some comments about possible activities of the
Sloan Foundation in support of interdisciplinary studies in
cognition.
.item←0

	#. Artificial intelligence's most important
need is an increased emphasis on basic research.  Besides AI's
internal tendency to attempt premature applications, there has
been a Government, especially DoD, tendency to demand short range
goals.

	#. One program that could help basic research immediately
would be a substantial number of nationally competitive postdoctoral
fellowship positions in cognitive science.  Much of the research
that needs doing is perhaps best done by young scientists without
other commitments and can be done on an individual basis.  Experimental
research in AI and computational linguistics requires computer
facilities, but these are available in a number of places.
Such a program is especially suited for a temporary effort such
as is planned by the Sloan Foundation.  Three year fellowships
seem the most appropriate.

	#. It would be worthwhile to equip more computer science
and computational linguistics departments with suitable computer
facilities.  Ten years ago, facilities suitable for AI work
cost more than $1,000,000, but now $200,000 will buy an excellent
facility for a moderate sized research group.  Some matching
might be expected.

	#. Projects intermediate in size between the individual
grants favored by NSF and the large ARPA supported laboratories
are viable and likely to be very productive.

	#. Endowed professorships, while not very exciting
to a foundation, put a solid basis under a field.

	Concerning the controversy with the Chomsky school, I have
only the following comments:
.item←0

	#. Their doctrinal insistence on the primacy of syntax
has always been harmful, although their specific contributions
to syntax have been useful.
Barbara Partee, whose knowledge I respect, tells me
that this doctrine has been modified recently, but
I gather it hasn't reached the stage of admitting that either
the AI approach or Montague's philosophical approach have made
contributions to the study of language.

	#. From my own point of view, and from the point of view
of everyone else in artificial intelligence, the most important
questions about language concern its semantics - what information
is conveyed by discourse.  I think that language is secondary
to thought; a linguistic utterance, especially a complicated one,
is the result of thinking about how to express one's thoughts in
a publically understandable way.  AI work confirms this view,
because this is the problem faced by a would-be robot designer,
and the experiments on teaching apes to communicate confirm it,
because apes, who do not ordinarily communicate facts to one
another, can be taught to do so.

	#. I have an aversion to involving funding agencies in
scientific controversies.  It heats up the arguments, wastes time,
and there is no guarantee that the best work will win.  Therefore,
in the face of controversy, solutions that allow young scientists
to vote with their feet, such as fellowship programs, have
much to recommend them.
However, I doubt that this controversy is either so fierce or
so pervasive as to require confining support to such programs.

	The above should be taken as a somewhat casual statement
of opinion and not as an attempt at a definitive presentation
of the "AI point of view".

.sgn