The Singularity is Near

Kurzweil's latest opus

Through a combination of good luck and low cunning, I've managed to get my sticky little fingers on an early copy of Kurzweil's latest book (many thanks to SIAI and AC2005 for providing my copies!)

The bottom line: it's everything I'd hoped for. Just as Spiritual Machines was pretty much an updated replacement for Intelligent Machines, The Singularity Is Near pretty much updates and replaces Spiritual Machines. Here, in no particular order, are my impressions from reading it:

The book opens with countless pages of Ray's infamous logarithmic charts, showing exponential growth in just about everthing. In addition to the usual suspects, he covers concepts like the time between the introduction of a new invention and its adoption by 25% of the US public, and per-capita wealth growth. Although I think Ray's belief in his "Law of Accelerating Returns" sometimes veers a little close to the mystical, he makes a very compelling case that we're entering the interesting part of a runaway information curve. From my perspective, the most relevant part of this section had to do with Moore's Law. Unlike Spiritual Machines, this book was able to convince me that Moore's Law is good for at least another decade before we need a new technology to replace it.

Ray also spends a lot of time talking about the human brain, and the progress that we're making in reverse-engineering it. In particular, he points out that most of the relevant technologies (such as brain scanners) are now information technologies, and are consequently improving at an exponential rate. From my perspective, the most interesting insight in this part is his claim that several parts of the brain have been reverse-engineered in detail, and that in each case, the computational power required to duplicate their functionality is about 1/1000th of the theoretical power of the corresponding portion of the brain.

A good chunk of the book is devoted to discussing the technologies that will lead up to and follow from the singularity (in particular, GNR: genetic engineering, nanotech and robotics (ie, AI)). This section of the book is very solid -- although he gets a little random at times, he's pretty good about noting what he's confident about and what's more speculative. I disagree with Ray about some of this -- I think that the singularity will happen considerably sooner than he does, and I expect a much faster "hard takeoff" than he does. Nonetheless, he presents a compelling and well thought-out timeline here.

Perhaps the weakest section of the book covers "existential risks" (ie, risks that stand a chance of wiping out the human race). It's a profoundly important section, and he covers most of the key points, but the presentation felt a little disjointed and less extensively researched to me. Incidentally, at AC2005, Michael Anissimov and I spent a little time conducting an informal poll on existential risk. The consensus (which I agree with) was that we stand something like a 50% chance of screwing up and wiping ourselves out over the next couple of decades. So it goes...

Whether or not you agree with him about everything, Kurzweil is clearly a formidable intellect backed by a very strong team of researchers. Like its predecessor, I expect that Singularity will be required reading for anyone who's remotely serious about futurism.