On Intelligence

Why hasn't Artificial Intelligence taken off? Why don't we have robots? How "smart" is your car? What is "intelligence" after all? Where does creativity enter the equation? Is the brain just a computer without the silicon?

Jeff Hawkins' book On Intelligence (written with co-author Sandra Blakeslee) addresses these questions and does an admirable job. Hawkins is the original architect of the Palm computer; Blakeslee is a 30-year veteran science writer for the NY Times.

The gist of the book is Hawkins' personal theory of intelligence and why AI has never gone anywhere, and will continue to go nowhere, until we change our way of thinking about Intelligence. He says this isn't a completely new theory but apparently the bits and pieces haven't been presented in a coherent and persuasive whole before.

The book is written very clearly in a nice, readable, conversational style. Although I suspect that much of the wordsmithing is Ms. Blakeslees, the ideas are all from Jeff Hawkins, representing thoughts and theories he's been formulating since 1979.

Jeff's theory is that AI hasn't worked (and won't work) because computer scientists dismiss the brain and how it works, deciding that "we can do better". Many AI "experts" insist that AI is just around the corner -- AI will be great "just as soon as we have faster more powerful computers".

Jeff Hawkins believes that before we can create 'intelligent machines" we have to understand what intelligence means and to do that we have to understand how the brain works. Well, duh.

He then presents quite a believable theory (at least, Rich and I both think so! :^)

On Intelligence is a very interesting and highly approachable book. Around chapter 3 or 4 I realized that Hawkins' theory of the brain does a good job of explaining how a friend of mine thinks. Further into the book I found several sections that fit well with the way I think.

These realizations makes Hawkins' theory a truly great theory because it fits the observed facts so well. I was amazed. I convinced my friend to read the book and he's also been finding many parts of it that match his life and the way he thinks (or so he thinks. :-)

Not everyone has been quite as taken with the book as Rich and I have been. One friend kept asking questions about emotions and "well-traveled paths of research" (until I advised him to read the book instead of the reviews. :-)

Another friend thought the first 4 chapters contained too much background information for his taste. My guess is that this is because most of his reading is of scientific and computer theory papers — papers which, almost by definition, don't contain much background at all. He's not used to it. I, on the other hand, found the willingness and ability of the authors to include approachable background material, bringing the reader up to speed on a fairly technical idea, to be well accomplished and admirable.

In any case, Rich and I have been recommending this book far and wide. Check your local library if not your local bookstore (Barnes & Noble has the hardcover version listed for only $15.) You may also be interested in the associated web site and discussion forum.