Back to the futurist

February 23, 2011

For 30 years Ray Kurzweil has been preaching the artificial intelligence gospel. As computers drive cars and play “Jeopardy,” are we on track to reach his cybertopia?

Andy Greenberg: You’ve predicted that by 2045 accelerating progress in technology means we’ll build a computer — the so-called singularity — powerful enough to let us upload and replicate our brains, essentially becoming immortal. Are we on schedule?

Ray Kurzweil: With regard to artificial intelligence, we’re very much on schedule. Just look at the actual tasks that AI is performing that weren’t possible three years ago. Driving a car in a busy urban situation with no driver: Google’s driverless cars have logged 140,000 miles. IBM’s “Jeopardy!”-playing supercomputer Watson plays at about the level of the best human players. I predicted in the early ’80s that a computer would defeat the world chess champion by 1998. It happened in ’97. Now people say, “Of course computers can play chess. But they’ll never be able to do something as human as understand language.” Watson is harder to dismiss.

AG: You also predicted we’d have a computer capable of 20 petaflops of processing by 2009, and we’re still less than a tenth of the way there.

RK: That computer is being built now. In 2011 there will be a 20 petaflop computer, the IBM Sequoia. So on that point we’re slightly behind schedule. […]