Computer Passes the Turing Test - Computer Program Surpasses Test's Threshold for the First Time
By Daniel Joseph Cruz | June 10, 2014 6:12 PM EST
Princeton University's computer program called Eugene Goostman achieved another accomplishment for passing and winning the official Turing Test at the Royal Society on Saturday, June 7. The Turing Test is an event organised by the University of Reading. It checks whether people can determined if they are communicating with a live person or a machine.
The test is based on Alan Turing's question-and-answer game "Can Machines Think?" Analysts have been using it for 65 years now. A computer program will pass the Turing test if users misidentify the program for a live human more than 30 percent during a five-minute chat conversation.
Princeton University's Eugene Goostman competed with other Artificial Intelligence systems like Cleverbot, Ultra Hal, and others. But Eugene won the competition as it scored 33 percent from the judges at the Royal Society in London. With this result, Eugene was first to surpass the 30 percent Turing Test threshold. Princeton University's Eugene program has already won the Turing Test two years ago with a 29 percent score.
Eugene Goostman is a computer program that simulates a 13-year old Ukrainian boy "turned into a chatter-bot by his school computer teachers," according the computer program's official testing Web site (follow this link). Eugene was created by Russian-born Vladimir Veselov and Ukranian-born Eugene Demchenko.
According to the University of Reading, the programmers were not given any of the questionnaires prior to the test and the judges and human control groups were kept apart during the test.
According to the organisers, Eugene's win was a great achievement, as no computer has passed the test with such a score before. There had been Turing Test-format competitions worldwide but the one held at Royal Society in London "was independently verified and, crucially, the conversations were unrestricted," according to Kevin Warwick, professor at the University of Reading, as reported by BBC.
Tech experts say people nowadays already have intelligent voice assistants in the form of Apple's Siri, Windows Phone's Cortana, and others that reply to its users' normal questions. But these mobile programs weren't designed to carry on a five-minute long conversation compared to Eugene and the other AI programs that recently competed at the Turing Test.
To contact the editor, e-mail: