A review by yates9
Superintelligence: Paths, Dangers, Strategies by Nick Bostrom

5.0

A technical philosophy book that deconstructs the arguments around use and development of AI as it develops to solve more general problems. This is the starting key reference for thinking about general intelligence and it goes very far toward being a complete analysis of the subject. Many of the assumptions we make around AI safety are debunked with this text, but ideas for alternatives are discussed, many new areas of research identified.
The only downside to the text other than density of content is that there are many underlying assumptions made to the discussion that are not always clear. Some of these, I disagree with so this text is an incomplete reflection of what I believe.