Superintelligence: Paths, Dangers, Strategies

This is more or less the technical tour de force of artificial intelligence books currently out there. If you're looking for a light introduction to the concepts of artificial intelligence, this is not the book for you.

Nick Bostrom goes in deep into the hurdles, risks and hazards of artificial intelligence. While I suspect there are academic books that go even deeper into the math and algorithms covered, Bostrom doesn't shy away from that discussion. It feels exhaustive. It feels like a technical textbook in many ways. The material is very dense. In addition to technical discussion of approaches to achieving AI and the challenges involved with them, there is deep discussion of the risks and even the ethics of AI. In my opinion, these were the most interesting parts.

Having read quite a few other books on the subject, I felt familiar with most of the terms used, as well as many of the major concepts. Even with this, this was a much more technical discussion than anything I'd read previously. With this in mind, I'll say I didn't really “enjoy” this book. But I do feel much better educated. I'm not sure how to account for this in “my rating.” I'm going to give it a 4 because this feels like the definitive discussion of the topic.