Hall looks at some of the history of AI, showing how creating machines to imitate humans often turns out to be much harder than expected. But computer power is increasing all of the time, and he argues that sooner or later computers will have a level of ability exceeding that of a human. He looks at some of the tasks that need to be mastered by such a computer, such as understanding human language, and discusses some of the ways that this might be implemented - does everything need to be put into the program at the start, or could the program learn in the same way as we do? Hall goes on to look at what the future is likely to bring, and at how we will fit in when machines are able to run everything for us (or for them). The book concludes with a discussion of how we might give artificial intelligences a moral sense, and at what status we should ascribe to them when we do so.
There's a lot of material in this book, but I found it to be something of a jumble, although it isn't particularly technical. I wouldn't recommend it to those wanting a straightforward overview of the possible futures of AI. It would be of more benefit to those who are interested in the subject and want to tap into Hall's extensive mine of ideas.