n the 1950 essay entitled "Computing
Machinery and Intelligence", Alan Turing proposed a test to determine if
machines could think whereby a computer tries to trick a person to think he/she was
chatting with a person via the computer.
The idea behind the test is that if a machine could
make a person think he/she was interacting with an intelligent person, why not consider
the machine intelligent in its own right? The Turing Test would be at least the best
approximation of intelligence possible.
It is this famous test that initiated the Loebner Prize contest in 1990. Philanthropist Hugh Loebner agreed with The
Cambridge Studies for Behavioral Science to sponsor $100,000 for the first computer to
convincingly pass the Turing Test by its response that is indistinguishable from a real
person. So far, no computer can communicate like an intelligent person so $2,000 is
awarded every year to the computer that seemed the most human.
The Turing Test does not lack its share of controversy.
Believers of weak AI and their more extreme counterparts argue that true human
intelligence will never be recreated in a machine. Other people feel that the Turing
Test side-steps the issue of what intelligence really is; therefore any computer that does
pass the test lacks some merit of seriousness at the very least. In a web page
entitled "Misguided
Artificial Intelligence: The Turing Test" author Vince Valler at Michigan State
University is against the Turing Test as the determiner of intelligence because it is
merely the simplification of true intelligence through a game designed to use tricks to
fool people to think it is truly intelligent. Citing the conversational computer
program Eliza as a primary example of tricks that can
fool people as well as other more sophisticated tricks used in other conversational
programs like Parry and Chatterbox, Valler concludes the Turing Test as a
"misdirection" just as the early alchemists sought to make gold out of metal.
The computer will never truly understand what is said to it or what it replies
because it is just manipulating and matching up appropriate symbols to appear intelligent,
as philosopher John Searle's "Chinese Room
Argument" contends as well.
Whether or not the Turing Test is truly a valid way to
determine machine intelligence is not likely to be resolved. For now, a machine that
can pass the test is not likely to be created any time soon, but as author Raymond
Kurzweil predicts, achieving a passing grade in the Turing Test will happen not too far
off from Turing's original prediction of 70 years.
