image

book review |

publication:
story:
date: January 2013

read | story

presented by

card :: the Seattle Times

book review |

Any book that promises on the cover to reveal “the secret of human thought” is setting the bar pretty high. But if anyone could be expected to pull it off, it’s inventor/futurist/Big Thinker Kurzweil, who’s been ruminating about the possibilities of human/computer convergence for more than 20 years now.

Kurzweil’s new book continues and expands on many of the themes he’s written about in earlier works such as “The Age of Intelligent Machines” (1990) and “The Singularity is Near” (2005). This time, though, Kurzweil also attempts to unravel the architecture of what until recently has been the most powerful computer in existence — the human brain — and how a conscious mind can arise from electrical charges caroming around a three-pound blob of gray matter.

His theory, put simply, is that the core job of the brain is recognizing patterns and organizing them into hierarchies of knowledge. He proposes that the neocortex is organized into roughly 300 million “pattern recognition modules” of about 100 neurons each. As we live, learn and experience the world, these modules forge connections with each other.

Critically, those connections are orderly, not the spaghetti bowllike tangle one might expect. Kurzweil cites neuroscientist Van J. Wedeen’s description: “Basically, the overall structure of the brain ends up resembling Manhattan, where you have a 2-D plan of streets and a third axis, an elevator going in the third dimension.”

Kurzweil then goes into great — too great, I fear — detail about how to go about building a digital brain that could replicate the thought processes of the biological brain. This discussion occupies most of the first half of the book, and far too often it has all the pizazz and narrative drive of an engineering manual.

This is a shame, because Kurzweil has a lot of perceptive things to say about the nature of human and machine thought, often drawing on his own experiments and inventions. But he gets bogged down in talk of hierarchical hidden Markov models and vector quantization methods; diagrams that look like they came from a computer-programming textbook don’t help much.

Fortunately, Kurzweil comes up for air in the last 100 pages or so. His description of how Watson, the Jeopardy! champion computer, actually worked is eye-opening and refreshingly clear; his consideration of the moral and ethical implications of intelligent machines makes it apparent that the human race will have a lot of adjustments to make if trends play out anywhere close to the way Kurzweil thinks they will.

A few other quibbles: Kurzweil likes to talk about himself, his work and his previous books, and while these references usually are on point their sheer number does get to seem a bit self-promotional. And in a (mercifully brief) late chapter, in which Kurzweil responds to critics of his past writings (among them our own Paul Allen), he sounds petty and thin-skinned.

Those don’t detract from the overall merit of the book. Whether or not you agree with his theories of how the brain works or the inevitability of us augmenting our brains with digital aids (now where is my iPhone…?) Kurzweil has thought seriously about where brain research and computer science are leading us, and what will happen when they converge. His book should prompt serious, intelligent discussion — which at least for now remains a strictly human prerogative.