XD

RUTGERS

THE STATE UNIVERSITY
OF NEW JERSEY

DEPARTMENT OF COMPUTER SCIENCE* HILL CENTER FOR THE MATHEMATICAL SCIENCES
BUSCH CAMPUS « NEW BRUNSWICK e NEW JERSEY 08903 « (201) 932-3546/2001

November 11, 1977

Dr. Joshua Lederberg

Department of Genetics, 5331
Stanford University Medical Center
Stanford, California 94305

Dear Josh,

In his message of Nov. 8, 1977, Elliott requested that comments about the Miller/Rand
proposal be sent to you. Here are my comments.

The ideas about a computer system to adjudicate claims for reimbursement are interesting
from the point of view of AT in legal reasoning. This is a new area of activity in the
AI field and it deserves support. It does have conceptual points of contact with diag-
nostic/interpretive reasoning but I find the relevance to AIM somewhat weak--especially
at the present state of development of the project.

 

The Rand work on information processing psychology seems to be mainly oriented to the
relationship between perceptual characteristics of linguistic stimuli and their concep-
tual processing. This is in the same general category as the U. of Pittsburgh research
by Lesgold and Greeno. Unlike the Lesgold/Greeno work however, which has already started
to use Anderson's ACT, the Rand work is intending to use a variety of fairly general
techniques that are being used in other SUMEX-AIM projects (e.g. production rules). I
find the argument for collaboration in the framework of AIM stronger in the Lesgold/Greeno
case then in the Rand case.

I am uncomfortable with Jim Miller's request for the use of SUMEX-AIM to complete his
dissertation work. Clearly, the main motivation is the difficulty with local computing
at UCLA. This creates a bad precedent for admission to the resource. The request
would be reasonable as part of a more general Rand program of participation in AIM. We
do not have as yet such a program; and I don't find Hayes-Roth's document convincing and
specific enough as a preliminary proposal for such a program.

In summary, I don't recommend the Miller/Rand project as a AIM-pilot project.

Sincerely , -
Swat a" iQ

Saul Amarel
Professor and
Chairman

SA/kb