Of course there are pockets of good work, like in any research institute. It's difficult to point to anything really pushing the envelope in AI research though. You have any evidence to the contrary?