SAINT

The invention:

Taking its name from the acronym for symbolic automatic
integrator, SAINT is recognized as the first “expert system”—
a computer program designed to perform mental tasks requiring
human expertise.

The person behind the invention:

James R. Slagle (1934-1994), an American computer scientist




 The Advent of Artificial Intelligence

In 1944, the Harvard-IBM Mark I was completed. This was an
electromechanical (that is, not fully electronic) digital computer
that was operated by means of coding instructions punched into
paper tape. The machine took about six seconds to perform a multiplication
operation, twelve for a division operation. In the following
year, 1945, the world’s first fully electronic digital computer,
the Electronic Numerical Integrator and Calculator (ENIAC),
became operational. This machine, which was constructed at the
University of Pennsylvania, was thirty meters long, three meters
high, and one meter deep.
At the same time that these machines were being built, a similar
machine was being constructed in the United Kingdom: the automated
computing engine (ACE).Akey figure in the British development
was Alan Turing, a mathematician who had used computers
to break German codes during World War II. After the war, Turing
became interested in the area of “computing machinery and intelligence.”
He posed the question “Can machines think?” and set the
following problem, which is known as the “Turing test.” This test
involves an interrogator who sits at a computer terminal and asks
questions on the terminal about a subject for which he or she seeks intelligent
answers. The interrogator does not know, however, whether
the system is linked to a human or if the responses are, in fact, generated
by a program that is acting intelligently. If the interrogator cannot
tell the difference between the human operator and the computer
system, then the system is said to have passed the Turing test
and has exhibited intelligent behavior.


SAINT: An Expert System


In the attempt to answer Turing’s question and create machines
that could pass the Turing test, researchers investigated techniques
for performing tasks that were considered to require expert levels of
knowledge. These tasks included games such as checkers, chess, and
poker. These games were chosen because the total possible number of
variations in each game was very large. This led the researchers to
several interesting questions for study. How do experts make a decision
in a particular set of circumstances? How can a problem such as
a game of chess be represented in terms of a computer program? Is it
possible to know why the system chose a particular solution?
One researcher, James R. Slagle at the Massachusetts Institute of
Technology, chose to develop a program that would be able to solve
elementary symbolic integration problems (involving the manipulation
of integrals in calculus) at the level of a good college freshman.
The program that Slagle constructed was known as SAINT, an
acronym for symbolic automatic integrator, and it is acknowledged
as the first “expert system.”
An expert system is a system that performs at the level of a human
expert. An expert system has three basic components: a knowledge
base, in which domain-specific information is held (for example, rules
on how best to perform certain types of integration problems); an inference
engine, which decides how to break down a given problem utilizing
the rules in the knowledge base; and a human-computer interface
that inputs data—in this case, the integral to be solved—and
outputs the result of performing the integration. Another feature of expert
systems is their ability to explain their reasoning.
The integration problems that could be solved by SAINT were
in the form of elementary integral functions. SAINT could perform
indefinite integration (also called “antidifferentiation”) on these
functions. In addition, it was capable of performing definite and
indefinite integration on trivial extensions of indefinite integration.
SAINT was tested on a set of eighty-six problems, fifty-four of
which were drawn from the MIT final examinations in freshman
calculus; it succeeded in solving all but two. Slagle added more
rules to the knowledge base so that problems of the type it encountered
but could not solve could be solved in the future.
   The power of the SAINT system was, in part, based on its ability
to perform integration through the adoption of a “heuristic” processing
system.Aheuristic method is one that helps in discovering a
problem’s solution by making plausible but feasible guesses about
the best strategy to apply next to the current problem situation. A
heuristic is a rule of thumb that makes it possible to take short cuts
in reaching a solution, rather than having to go through every step
in a solution path. These heuristic rules are contained in the knowledge
base. SAINT was written in the LISP programming language
and ran on an IBM 7090 computer. The program and research were
Slagle’s doctoral dissertation.

 Consequences

 The SAINT system that Slagle developed was significant for several
reasons: First, it was the first serious attempt at producing a
program that could come close to passing the Turing test. Second, it
brought the idea of representing an expert’s knowledge in a computer
program together with strategies for solving complex and difficult
problems in an area that previously required human expertise.
Third, it identified the area of knowledge-based systems and
 showed that computers could feasibly be used for programs that
did not relate to business data processing. Fourth, the SAINT system
showed how the use of heuristic rules and information could
lead to the solution of problems that could not have been solved
previously because of the amount of time needed to calculate a solution.
SAINT’s major impact was in outlining the uses of these techniques,
which led to continued research in the subfield of artificial
intelligence that became known as expert systems.


  
James R. Slagle

James R. Slagle was born in 1934 in Brooklyn,NewYork, and
attended nearby St. John’s University. He majored in mathematics
and graduated with a bachelor of science degree in 1955,
also winning the highest scholastic average award. While earning
his master’s degree (1957) and doctorate (1961) at the Massachusetts
Institute of Technology (MIT), he was a staff mathematician
in the university’s Lincoln Laboratory.
Slagle taught in MIT’s electrical engineering department
part-time after completing his dissertation on the first expert
computer system and then moved to Lawrence-Livermore
National Laboratory near Berkeley, California. While working
there he also taught at the University of California. From 1967
until 1974 he was an adjunct member of the computer science
faculty of Johns Hopkins University in Baltimore, Maryland,
and then was appointed chief of the computer science laboratory
at the Naval Research Laboratory (NRL) inWashington, D.C., receiving
the Outstanding Handicapped Federal Employee of the
Year Award in 1979. In 1984 he was made a special assistant in
the Navy Center for Applied Research in Artificial Intelligence
at NRL but left in 1984 to become Distinguished Professor of
Computer Science at the University of Minnesota.
In these various positions Slagle helped mature the fledgling
discipline of artificial intelligence, publishing the influential
book Artificial Intelligence in 1971. He developed an expert system
designed to set up other expert systems—A Generalized
Network-based Expert System Shell, or AGNESS. He also worked
on parallel expert systems, artificial neural networks, timebased
logic, and methods for uncovering causal knowledge in
large databases. He died in 1994.