Life 3.0

Life 3.0: Being Human in the Age of Artificial Intelligence by Max Tegmark takes you on a journey through a possible future we might inhabit with our intelligent friends. Those friends may not have flesh and blood, they may make the whole universe a computational machine, and if we will be around to enjoy it, who knows.

Tegmark's Life 1.0 cannot change its software or hardware except through random mutations, as with bacteria. Life 2.0, such as animals, can change its software; Life 3.0 can change both. I like that this classification leaves room for AI to take the third spot, and perhaps for humans, as Life 2.1, to join it.

He explores a wide range of AI scenarios. Some would be positive from our perspective; others would turn the universe into a paperclip machine. Alongside the fictional story, I enjoyed his discussion of the alignment problem (see Superintelligence by Nick Bostrom) and his account of the growing field of AI safety research, which also connects to Effective Altruism. Intelligence, in his framing, is independent of its physical form: it might be made of bits and bytes or of neurons and synapses.

Our technology gives fewer people the power to do more good and more harm. I hope we avoid an extinction event in the next 100 years. At the time I read the book, AI seemed narrow and was beginning to automate jobs through robotics. I pictured jobs as a hilly landscape, with AI moving from its lower slopes upwards. A surgeon, for example, might be worse than an AI at interpreting scans yet better at explaining the result to a family. I expect many jobs to disappear, which raises questions about how we organise society. Tegmark gives this less space, though I agree with his point that meaningful activity can exist beyond work.

One line I saved from the book: “Intelligence is simply the ability to solve complex tasks”. See the Goodreads book page.