On February 14th through 16th, the reigning champions from Jeopardy! are returning to take on a formidable new competitor -- a supercomputer. IBM's “Watson” supercomputer -- under development for four years -- will compete against Ken Jennings and Brad Rutter.
In a new interview with USA Today, he observed how a computer, IBM's Deep Blue, beat the world's reigning chess champion, Gary Kasparov, in 1997. However, for a computer, chess is child's play compared to the challenge of a game show such as Jeopardy, Kurzweil says. "This threshold is going to be harder to dismiss," he explains. "What people have emphasized, and I've agreed, that the key to human intelligence is really mastering the subtleties of human language -- things like puns and jokes and metaphors. If you look at the queries in Jeopardy, they’re really quite complex and subtle, and exactly what they’re talking about is not so clear. Watson appears to be able to get it." (Watson already won the practice round.)
Even if Watson doesn't win this upcoming match, "it will come close. And it will come back and win in the very near future, because it's only going to get better. And humans are not getting better."
Kurzweil calls the Watson Jeopardy match-up a "milestone" in the progression of machines to achieving human intelligence, which will be reached within 20 years, he believes. But people shouldn't fret about a machine invasion, he adds. "We're creating them to make ourselves smarter. We're going to literally merge with them."
By 2045, Kurzweil predicts, "we're going to multiply the current human intelligence a billion-fold -- which is really not so fantastic when you consider how far computers have come already."