A computer possessing human intelligence has been a thought dreamed up in the minds of prominent thinkers for some time. Alan Turing set the gold standard for such a computer in the late 1940s with his famous test; a test which was passed with ease by the fictional computer Hal in the film 2001: A Space Odyssey.
Such computers have remained in the fictional world until now. But in 2011 something amazing happened in America when IBM set up its computer "Watson" against Ken Jennings in the famous quiz show "Jeopardy". Jennings was the "winningest" ever competitor of the game, which demands a sophisticated and wide ranging field of general knowledge. In addition, the format requires a special skill. The questions are given in the form of answers, for example: "Built in the 11th century, it is a fortress built on the banks of the River Thames in England's capital city", for which the answer is given in the form of a question: "What is the Tower of London?"
Watson, loaded up with the entire content of Wikipedia, about 200 million pages of text, but not connected to the internet, had to give responses in real time in a human-sounding voice in five seconds or less (actually it usually supplied the answers in less than one second) . In Ken Jennings it was up against a phenomenally successful contestant who had previously set the record for consecutive wins in the show. What happened? Watson won, indeed Watson creamed its opponent and a couple of other high-achieving contestants.
Was Watson thinking? It is still a tricky issue, but we must concede it worked in such a way as to imitate human intelligence and surpassed the best humanity could put up against it. If you play jeopardy, presumably you are thinking. Is Watson thinking when it plays? It is beginning to be hard to say it isn't. And if so, I suspect we are very close to a Watson-like machine sailing through the Turing Test. And when that happens the implications will be huge...
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment