There is a great deal of good material here. I wonder if the general problem of producing a history of AI would not have been better decomposed into a set of mini-histories each concerned with a particular topic and occupying a single chapter. For example, NLP, machine vision, robotics, knowledge representation, vagaries of public/private financing, major commercial deployments, etc. each could have been addressed in a single chapter. In this book, individual topics pop up again and again in interleaved fashion as the author's single timeline unfolds. This may be a bit disconcerting for readers not already well versed in the field. Other advantages of a modular approach to the history of AI rather than a simple sequential approach are ease of updating the text for future editions and the ability of subject matter experts to quickly find and provide constructive feedback in their areas of expertise.



A minor irritation was the use of URLs in body text rather than confining them to end notes. Most authors would like their books to be timeless; the use of highly fragile URLs in body text seems to contradict this goal.



I suspect that this is the best history of AI we have so far. I recommend this book to anyone interested in the field.