From Alan Turing to Arthur C. Clarke, visionaries have dreamed of computers that could ape their human masters. But 2001 came and went, and a couple of decades on HAL is no closer. Instead the future of AI suggests machines that are very powerful but think differently from us. Must we accept that our human version of intelligence is only one of many, leaving us no special grasp on the universe? Or are all versions of thought essentially the same, and a new HAL just around the corner?