On February 11th, 2019, President Trump signed an executive order launching the "American A.I. Initiative" (carrying on from previous White House announcements on A.I.). A White House spokesperson has said the initiative outlines "bold, decisive actions to ensure that AI continues to be fuelled by American ingenuity, reflects American values and is applied for the benefit of the American people."
As you'll read in my upcoming book "Rage Inside the Machine: The Prejudice of Algorithms, and How to Stop the Internet Making Bigots of Us All," algorithms do indeed have values of their own. And we should all be very concerned about A.I.s adopting what the current US Administration thinks are "American Values."
This concern should come from the fact that algorithms are far too easily tied to prejudices against people. Pre-judging explicitly requires the reduction of people to categories and quantities, and as the book discusses, the scientific techniques that have traditionally been tied to simplifying people in this way are the direct ancestors of techniques in the algorithms (and thus A.I.s) of today.
With appropriate human caution, these techniques have been a valuable part of social science, but they have also often drifted into supporting prejudice, intolerance, and hate.
The book is about how appropriate human caution in the use of A.I. is now more critical than ever, and how that caution can take technical form, in creating algorithms that promote true diversity instead of polarization, adaptabilty rather than rigid and simple-minded optimization.
The book is available for pre-order from Amazon now.