The AI Does Not Hate You: Superintelligence, Rationality and the Race to Save the World

When Artificial Intelligence is depicted in films and books it tends to be capable of learning to feel emotions, and absorbing moral values. But, says Tom Chivers, true AI will be nothing like we imagine. The problem is that we are the only general intelligence (a general intelligence is an intelligence that can be applied to any problem) that we have ever encountered, so we inevitably picture AI through an anthropomorphic lens. We're very wrong to do so. We evolved in an unconscious and contingent way; there's no reason to suppose that an artificial brain would be anything like a human one. AI might be exceptionally intelligent (perhaps more intelligent than us by the same degree of magnitude as we are more intelligent than nematode worms) and value almost anything. Even, as one particularly brutal and apocalyptic thought experiment reveals, paperclips.

From the origins of AI in Turing's codebreaking group at the end of WW2, to Deep Blue, AlphaGo, and the technological singularity, Tom Chivers' book will be an examination of AI through the lens of the Rationalists: the community of coders, scientists and bloggers on the West Coast who are making a sustained attempt to understand how thought (and therefore AI) works. It will be a work of popular science and philosophy, as well as a Jon Ronson-esque deep dive into a very strange and occasionally sinister internet subculture.