I had an hour to spend last night, and used it to leaf through Ray Kurzweil’s new book, The Singularity is Near, in which he argues that by (roughly) 2045, computer intelligence (or, at least, processing capacity) will be bigger than all human brains combined. This will lead to a merger of technological and human intelligence, and, in time, to the "awakening of the universe" – which I understood to be a sort of mobilization of every molecule in the universe in the service of creating intelligence.
Ray Kurzweil uses many exponential graphs to make his argument, which he sums up as the world going through six epochs (see figure) – physics, biology, brains, technology, merger of technology and humans, and, lastly, the awakening.
I don’t know. Kurzweil has a great track record on predictions with his previous books, and certainly knows how to provoke. Whenever I want to irritate my students, I give them Alan Turings Computing Machinery and Intelligence and a couple of chapters from Kurzweil’s The Age of Spiritual Machines. The following discussion is always interesting, especially when students have to come to terms with what "intelligent" means.
But I can’t help feeling that there is some sort of Achilles and the tortoise about the singularity argument – I cannot bring myself to understand what happens as all these trends converge and their growth approaches infinity – nor that they necessarily do. I have always argued that processing and communication should be thought of as free resources, but, of course, within overviewable limits.
The upshot, of course, is that there is not need to panic – I’ll be 84 in 2045, and even if I don’t make it there (though, if Kurzweil is right, we all stand a pretty good chance of getting there and much longer) I will adopt a real options strategy, which is, I will wait and see, and not worry too much about it. We will know soon enough.
(Incidentally, a lot of stuff is available on the singularity.com website, including chapters 1 (the six epochs) and chapter 9 (response to various criticisms).