Can computers ever be intelligent? Hollywood would like to think so. Ever since the
early 1960s, free thinking machines have entered the mainstream of Science-Fiction
films, from the evil "Hal" from 2001: A Space Odyssey to the elegant "Data" in Star T
to Turing's criterion.
In 1950, Alan Turing devised a test to determine intelligence of a digital computer in
his historic essay, Computing Machinery and Intelligence . His name for the test was
the "Imitation game," which was later named the "Turing Test" by members of the AI
test was held on November 8, 1991 in Boston's Computer Museum. The contest was
called the Loebner Contest, named after a business man Hugh Loebner who offered
a $100,000 dollar prize to the author of the first program to pass the full Turing test. In t
To this day, the AI community cannot agree on how it is we are intelligent. If we are
conscious, self-aware, understanding, rational beings, and we are also intelligent, are
we intelligent because we are conscious, self-aware, and rational, or are these
achine's outward behavior is indistinguishable from the intellectual behavior of a
human, then the machine is intelligent. Turing implies that what is happening within
the computer is irrelevant to the question of intelligence. This definition omits the The
definition of intelligence Turing proposed almost fifty years ago still remains a valid
one. Members of the AI community have accepted his definition as a law. Still others
refute his definition. In attempt to show that Turing's definition of intel 1. Store
2. Executive unit,
Like and infant growing to adulthood constantly taking in data and storing it, the
computer can also receive and store inputs. With technology today, this storage can
be nearly infinite. The executive unit in an infant is the infant's ability to access he
store, and dictates the computer's behavior.
It is clear that although there exist significant parallels between the broad functions of
a human mind and those of a computer, there seems to be a fundamental difference
between these two systems: The computer simply recalls information stored in its d
But wouldn't that just be an application of already known discoveries applied to a
different problem? If this type of work is defined as original, then a computer can
easily produce original work by linking information in its databases together applying
n thought. The only difference there seems to be is the lack of consciousness on the
part of the computer.
I would now like to take apart the argument of consciousness Turing addressed in his
paper with a modern example. The argument from consciousness is simple: In order
to know a machine thinks, one would have to somehow find out if the machine knows
it is s with Chinese characters on them. When a native Chinese speaker who acts as
a judge inserts a phrase by means of index cards through the slot, the man must
formulate a response. But the rule book does not have translations for the characters.
Instead, Searle states that no computer program could ever understand anything as
we understand things. Programs mimic the actions of the English speaker, they follow
rules to manipulate meaningless symbols. Although the output of the computer is
meaningful to u become the machine and experience the consciousness it is
The Turing test still remains the most accurate means of measuring intelligence. It is
clear that computers "think" differently than humans. Philosophers like John Searle
support the claim that computers will be able to think consciously, although not i
Epstein, Robert. The Quest for the Thinking Computer. AI Magazine, pages
Garner, Robby. The Idea of FRED, ALMA, Issue 1, January 18, 1996
Gribbin, John. In Search of Schroedinger's Cat. New York. 1984. p163.
Turing, A.M., 1950. Computing Machinery and Intelligence, Mind 59: 433-460.
Reprinted in: Haugeland, John. Mind Design II. 1997, 29-56
Plato, Meno. Indianapolis, Indiana. 1949. p44.
Searle, John. Minds, Brains and Programs. Behavior and Brain Sciences