The Best of Creative Computing Volume 2 (published 1977)
Primer on Artificial Intelligence (State-of-the-art. bibliography)
V. STATE-OF-THE-ART
Research in the field of artificial intelligence has yielded much fruit since
its infancy only twenty-five years ago. But the fruit it has borne is nothing
compared to what it must bear in the future if it is to realize the predictions
of its adherents.
Accomplishments
Artificial intelligence is a regenerative science; that is, accomplishments of
today rapidly build those of tomorrow. This can probably be said about any
field of endeavor, but nowhere is it more true than here. The reason being
that, as we progress, the computer becomes stronger and more capable of aiding
the researcher in his quest.
Research in artificial intelligence has given various byproducts such as
assembly programs, debugging programs, test editing programs, and even a good
mechanical arm developed at the Massachusetts Institute of Technology. There
have been programs written that compose music and some that find chemical
structures. In addition, there have been extremely complex programs written
which handle such diverse tasks as traffic control, aiding architectural design,
aiding electronic circuit design, monitoring patients in a hospital, and
simulating chemical reactions.
By definition, artificial intelligence must have the ability to adapt to its
environment and react to totally unforeseen circumstances. Although much has
been accomplished, almost all computer applications fail to qualify as
artificial intelligence under this definition.
Goals
The basic problem in artificial intelligence, as yet unresolved, is that all
alternatives must be made explicit. Ways must be found for programs to
determine useful information out of collections of seemingly random bits of
information.
Computer programs must have shortcuts, similar in nature to those used by the
human brain, built into them. In addition, they must be able to deal with facts
about objects, relations between objects, and facts about facts. Also, all
behaviors must be representable by the program, and it must be able to have or
evolve concepts of partial success.
Furthermore, creative purpose is absent in the computer. In the words of Donald
G. Fink: "At best, today's computers can only assist man in creative work." If
artificial intelligence is to become an actuality, then one of the highest goals
of the researcher must be to find ways to give machines creative purpose.
Computers are not constrained by a sociological environment; their only
limitations are imposed by the programs given to them. Therefore, we must be
sure that the 'top goal' given to a highly intelligent machine be the welfare of
humanity, and not some private goal of the machine. As David Kendall said in
the preface of a 1967 book of collected papers on machine intelligence: "Thus a
study of 'Machine Intelligence' leads after yet one more remove to the ancient
imperative, 'know thyself' and the universal problem of coexistence."
Predictions
There have been many predictions made in regard to the progress of artificial
intelligence research. Among these is the 1965 prediction by Simon that:
"…machines will be capable, within twenty years, of doing any work that a man
can do." On the other side of the coin is the statement by P. E. Greenwood:
"From the brief summary of the state of the art of artificial intelligence, one
would conclude that little progress has been made since about 1960 and the
prospects for the near future are not bright." Whichever side one decides to
take, there are ample adherents to each. Marvin Minsky sums it up nicely:
Once we have devised programs with a genuine capacity for self-improvement a
rapid evolutionary process will begin. As the machine improves both itself and
its model of itself, we shall begin to see all the phenomena associated with the
terms 'consciousness,' 'intuition' and 'intelligence' itself. It is hard to say
how close we are to this threshold, but once it is crossed the world will not be
the same.
VI. CONCLUSION
Artificial intelligence is a young science. Only time and the persistence of
artificial intelligence researchers will reveal the full measure of its
potential.
The late A.M. Turing laid down a test for adaptive intelligence. The test,
which has come to be known widely as the Turing Test, is very simple. It
consists of a man (the examiner) trying to discern whether or not the responses
to questions he has proposed are being answered by a computer or another man.
The test is conducted in such a way that the questioner can not physically
determine which respondent has produced the response to his question; he must
make the determination solely from the answer itself. So far, no one has been
able to write a program that can consistently fool the examiner in the Turing
Test. Until this is done, true artificial intelligence will remain within the
realm of science fiction.
BIBLIOGRAPHY
Arnold, Robert, Harold C. Hill and Aylmer V. Nichols. Modern Data Processing.
New York, New York: Wiley, 1969.
Bobrow, Daniel G. "Natural Language Input for a Computer Problem Solving
System." In Semantic Information Processing, edited by Marvin Minsky. Cambridge,
Massachusetts: M.I.T. Press, 1968, 146-226.
Cannara, Alexander B. "Toward a Human Computer Language." Creative Computing, 1
(September-October, 1975), 38-40.
Chu, Kenneth C. "Applications of Artificial Intelligence to Chemistry."
Analytical Chemistry, 46 (August, 1974), 1181 -1 1 87.
Collins, N.L. and Donald Michie (Editors). Machine Intelligence 1. New York, New
York: American Elsevier, 1967.
Dale, Ella, and Donald Michie (Editors). Machine Intelligence 2. New York, New
York: American Elsevier, 1968.
Dreyfus, Hubert L. What Computers Can't Do – A Critique of Artificial Reason.
New York, New York: Harper and Row, 1972.
Feigenbaum, Edward A. and Julian Feldman (Editors). Computers and Thought. New
York, New York: McGraw-Hill, 1963.
Fink, Donald G. Computers and the Human Mind. Garden City, New York: Doubleday,
1966.
Good, I.J, "A Five Year Plan for Automatic Chess." In Machine Intelligence 2,
edited by Ella Dale and Donald Michie. New York, New York: American Elsevier,
1968, 89-118.
Hamming, Richard W. Computers and Society. New York. New York: McGraw-Hill,
1972.
Kiss, G.R. "Networks as Models of Word Storage." In Machine Intelligence 1,
edited by N.L. Collins and Donald Michie. New York, New York: American Elsevier,
1967. 155-167.
McCarthy, John. "Programs with Common Sense." In Semantic Information
Processing, edited by Marvin Minsky. Cambridge, Massachusetts: M.I.T. Press,
1968, 403-418.
Minsky, Marvin (Editor). Semantic Information Processing. Cambridge,
Massachusetts: M.I.T. Press, 1968.
Newell, Allen and H.A. Simon. "GPS, a Program That Simulates Human Behavior." In
Computers and Thought, edited by Edward A. Feigenbaum and Julian Feldman. New
York, New York: McGraw-Hill, 1963, 279-293.
Porter, Arthur. Cybernetics Simplified. New York, New York: Barnes and Noble,
1969.
Quillan, Ross M. "Semantic Memory." In Semantic Information Processing, edited
by Marvin Minsky. Cambridge, Massachusetts: M.I.T. Press, 1968, 227-270.
Slagel, James R. Artificial intelligence – The Heuristic Programming Approach.
New York, New York: McGraw-Hill, 1971.
Streeter, Donald N. The Scientific Process and the Computer. New York, New York:
Wiley, 1974.
Wiest, Jerome D. "Heuristic Programs for Decision Making." Harvard Business
Review, 44 (September-October, 1966), 129-142.
