Hawkins on intelligence: Fascination and frustration

I find On Intelligence to be both fascinating and frustrating. Fascinating in that it hints at a potentially unifying view (the memory-prediction model) of brain structure and function; and frustrating in that it is very sketchy and in that it also makes claims about AI and conscious experience that seem to me to be questionable. I will mostly leave the neural aspects to other reviewers, focusing my remarks primarily on four things: (i) general comments about the model and how it relates to AI, (ii) Deep Blue as case study, (iii) the Turing Test and its impact on AI, and (iv) consciousness. But first some general comments on the book, to orient the reader. Jeff Hawkins (with co-author Sandra Blakeslee) has written an easy-to-read book on a topic of broad general interest—the nature of intelligence and how it is that brains have it and why machines don’t yet—in which his own excitement comes through and infects the reader as well. The account is largely biographical, in which Hawkins (inventor of the Palm Pilot) describes how he came to his current ideas about artificial intelligence, the brain, and related topics including Deep Blue, the Turing Test and the Chinese Room. He also describes the ideas themselves in an engaging way, and ends the whole thing with eleven in-principle testable theses that his theory suggests. The upshot is that he thinks the