ACKNOWLEDGMENTS

It would have been impossible to write How Smart Machines Think without the hard work of the many researchers and engineers behind the breakthroughs in the following pages, as well as the journalists who have covered many of the more “human” details of their efforts. In some sense, this book was easy for me to write, because these researchers have already done most of the hard work. They’ve spent thousands of person-years on experiments, research, and documenting their findings. My role in writing this book has primarily been to collate and organize the results of their research into a more digestible form.

Many people and organizations have been helpful in the development of this book. My family has provided invaluable support through the process. This includes my wife, Sarah, who read more drafts than she would have liked, and my parents and brothers for their role in introducing me to computers in the first place. Teza Technologies was very generous for finding a way to accommodate my desire to write this book (during nights and weekends) while working at the company. From Teza, Michael Tucker reviewed the entire manuscript and provided helpful feedback. MIT Press’s editorial staff, including Marie Lufkin Lee, Marcy Ross, and Christine Savage, have also been extremely collegial and helpful through the process, and anonymous reviewers who read early drafts provided helpful feedback. Mary Bagg additionally provided many helpful suggestions, comments, and corrections to the manuscript, and South Park Commons provided a nice community as I added the finishing touches to the book.

Many friends have also offered their ideas, time, and introductions, including Eric Jankowski, Andrew Cowitt, and Ricky Wong. Daniel Duckworth provided especially thorough feedback on the Jeopardy chapters, and my dad, Gary Gerrish, provided helpful feedback on the first half of the manuscript. David Churchill, Ben Weber, Jie Tang, James Fan, and Chris Volinsky have also graciously lent their time to answer questions about some of their work that I’ve discussed in this book, and to provide feedback on chapter drafts. Jason Yosinski and his colleagues also kindly allowed me to use their images of what neural networks see in chapter 9, the first full chapter about deep learning, and Alex Krizhevsky gave me permission to use an image of AlexNet (in the end I used a variant of that image).

Finally, I am fortunate to have had guidance from various mentors, formal or not, who have shown me many of these ideas throughout the years, before I began work on this book.