Where I read it: mostly on the Tube, and a little bit during lunch breaks
This was the first non-fiction book that I got to on my 20 Books of Summer list. As I think I mentioned in my June Superlatives round-up, I have almost no background in computer engineering, evolutionary biology, or mathematics, so it was, to say the least, harder going than any of the fiction I’d read so far. Fortunately, George Dyson is a solidly competent writer; confusion never arose because he was confusing, just because I often didn’t have the knowledge that would have clarified things for me. He also has a distinguished scientific pedigree: his mother was a famous mathematician, Verena Huber-Dyson, and his father was Freeman Dyson, theoretical physicist and inventor of the Dyson sphere. (More in my wheelhouse was his grandfather, the Edwardian composer Sir George Dyson, responsible for the Evensong canticle settings Dyson in F [aka the Star Wars Service] and Dyson in G. And some other stuff, too.)
Dyson’s thing is machine intelligence. This book is all about how, if and when (and it’s mostly when) machine intelligence arises, it’s likely to do so through processes similar to those that created life as we know it. Computers, in other words, are going to experience evolution, or rather,a version of natural selection. Conditions that are advantageous to a computer network will allow pieces of that network to flourish, until it’s able to respond and adapt to its own environment without any input from the engineers that built its circuits or the programmers that set it in motion.
This is the sort of thing that people (especially fiction writers) refer to as Artificial Intelligence, and AI bots already exist – they’re just not the kind of bots you really want to be hanging out with. Dyson is a science historian, though, not a fortune teller, so he focuses less on the possibilities and more on the history of the belief that humans will someday create a global intelligence. It’s older than you think. It predates Turing by centuries; Dyson pinpoints the beginning of the idea with Hobbes and Leviathan. Hobbes sees the government, the state, as a kind of collective entity composed of a nation’s people:
“Nature is by the Art of man, as in many other things, so in this also imitated, that it can make an Artificial Animal.”
He sees the Internet as a perfect place for a global intelligence to develop: a network that spans the world, through which vast amounts of data travel in fractions of a second. Here, also, he suggests that evolution, which is generally painted as a thoughtless or “randomized” process, might be driven by considerations that could be referred to as intelligent ones. It sails close to the wind of intelligent design, but not in the way that Texan fundamentalists think of it; rather, Dyson suggests that “intelligence” may be a concept we are applying all wrong. Machine intelligence may be something that already exists but which we are simply failing to recognize because it is so far above, beyond, and/or different to, the ways in which we understand human intelligence to work. It’s an argument that allows for the existence of something like a God, in the same way that you can call “magic” a kind of science we don’t yet understand. It’s perhaps the scariest, and yet the most beautiful, idea in the book. For all that I could have done with a greater depth of knowledge while reading it, I’m very glad I did.
(I’ve now passed it on to the Chaos, who will probably have more nuanced things to say.)
Darwin Among the Machines, George Dyson (London: Penguin, 2012 )