This weekend, to mark the 60th anniversary of Alan Turing's death, a chatbot named Eugene Goostman—a program pretending to be a 13-year-old Ukrainian boy—fooled one of three assembled judges into thinking that
he it is human. Whether this marks the first beating of the Turing Test, the pioneering computer scientist's trial for artificial intelligence, remains a matter of debate; for one thing, one of Turing's qualifications was that the human-fooling be done repeatedly. For another thing, though, there have been other programs that have claimed Turing Test passage.
And two of them, one time, talked to each other.
In 1966, the computer scientist Joseph Weizenbaum created a program that seemed to be a contender for Turing Test passage. The early chatbot searched for keywords in conversations conducted with human typers; if the human used one of those words, the program would use it in its reply. If not, it would offer a generic response. It was meant to mimic a psychotherapist (which allowed it to adopt "the pose of knowing almost nothing of the real world"); he named it ELIZA. And in 1972, the Stanford scientist Kenneth Colby created another program, PARRY—a bot that tried to model the behavior of a paranoid schizophrenic. PARRY was, Colby said, "ELIZA with attitude."