Kurzweil Is Confident Machines Will Pass Turing Test by 2029 (video)
It continues to be a good Spring for Ray Kurzweil. Not only has the media been rather kind to him lately, some of his biggest predictions about computer intelligence seem more and more likely. Kurzweil has famously predicted that machines will match human intelligence and pass the Turing Test by 2029. Seem far fetched? Well, IBM's Watson system just surpassed the Jeopardy! skills of Ken Jennings. That's a pretty good sign of things to come, at least according to Kurzweil. In a new video interview taken at the iLabs Singularity Summit in Italy, the noted futurist seems pretty happy with the recent trends in artificial intelligence, especially Watson's win and the human reaction to it. Check out David Orban's seven minute chat with Kurzweil in the video below. Could Watson's successors really pass the Turing Test in 20 years? Seems pretty likely to me, too.
Orban actually gets to the Watson phenomenon towards the end of the interview. Skip to 4:13 to see that discussion. Kurzweil explores Watson's chances with the Turing Test at 6:16. His famous Long Bet with Mitch Kapor, which we've discussed a bit before, comes up at 6:55. Don't forget to go back to the opening eventually, as Kurzweil gives some interesting commentary on how our language of technology doesn't accurately reflect its reality:
I share Kurzweil's admiration for Watson and its programmers. As he points out, the system has amazing language processing skills even if they aren't quite up to human levels. It also can assimilate vast stores of knowledge and retains that information in ways that meet or exceed the majority of humans. I differ form Kurzweil, however, in thinking that it was these skills that let Watson win Jeopardy! As Ken Jennings pointed out after the event, most Jeopardy! champions have the exact same skillset and the computer's trouncing of humans may actually be due to superior timing on the buzzer rather than superior knowledge. In any case, I can't argue that hearing Watson referred to as a 'he' time and again is pretty cool, and there's little doubt that the IBM system is a great sign that AI is actually making some steady progress.
Will it continue to grow fast enough to reach human levels of (emotional) intelligence and pass the Turing Test by 2029? I find it hard to argue against the notion. With Watson the language processing skills are getting closer, and the ability to assimilate mass amounts of data and synthesize an acceptable answer is clearly a great step in the right direction. It may take much longer than 20 years to get a computer intelligence to think in the same way as a human, but getting one to talk and converse like a human could happen much sooner. 2011 is shaping up to be a great year for Kurzweil, 2029 could be a great year for everyone.
[screen capture and video credit: David Orban]