There’s an earlier post here on this initiative at IBM, and the advancements continue. To me, this sort of natural-language research is the most intriguing area in artificial intelligence. It’s not so much the ability to come up with the correct response (difficult enough, since the “machine” is disconnected from the resources of the Internet); it’s the accurate parsing and interpretation of the cagey, pun-laden clues and categories that makes this really amazing.
