Turning Point: Policymaking in the Era of Artificial Intelligence

by Darrell M. West, John R. Allen

Artificial Intelligence is here, today. How can society make the best use of it? Until recently, “artificial intelligence” sounded like something out of science fiction. But the technology of artificial intelligence, AI, is becoming increasingly common, from self-driving cars to e-commerce algorithms that seem to know what you want to buy before you do. Throughout the economy and many aspects of daily life, artificial intelligence has become the transformative technology of our time. Despite its current and potential benefits, AI is little understood by the larger public and widely feared. The rapid growth of artificial intelligence has given rise to concerns that hidden technology will create a dystopian world of increased income inequality, a total lack of privacy, and perhaps a broad threat to humanity itself. In their compelling and readable book, two experts at Brookings discuss both the opportunities and risks posed by artificial intelligence—and how near-term policy decisions could determine whether the technology leads to utopia or dystopia. Drawing on in-depth studies of major uses of AI, the authors detail how the technology actually works. They outline a policy and governance blueprint for gaining the benefits of artificial intelligence while minimizing its potential downsides. The book offers major recommendations for actions that governments, businesses, and individuals can take to promote trustworthy and responsible artificial intelligence. Their recommendations creation of ethical principles, strengthening government oversight, defining corporate culpability, establishment of advisory boards at federal agencies, using third-party audits to reduce biases inherent in algorithms, tightening personal privacy requirements, using insurance to mitigate exposure to AI risks, broadening decision-making about AI uses and procedures, penalizing malicious uses of new technologies, and taking pro-active steps to address how artificial intelligence affects the workforce. Turning Point is essential reading for anyone concerned about how artificial intelligence works and what can be done to ensure its benefits outweigh its harm.

Genres: Technology, Artificial Intelligence, Nonfiction

297 pages, Hardcover

Published July 28, 2020

Reviews

[Groucho42 · Rating 5 out of 5]

An excellent introduction to policy issues for artificial intelligence. It takes a higher level view of AI overall, and delves in to the five sectors of healthcare, education, transportation, ecommerce and defense. While they miss a point about the importance of how to provide a regulatory environment that better manages deep learning as different from hardware and procedural code, the wide overview they provide is necessary to any executive or government official considering how policies need to adapt based on the rapidly widening adoption of AI in both the public and private sectors.

[Aketzle · Rating 3 out of 5]

I had high hopes for this book after hearing a really interesting interview on the subject with one of the authors, John Allen, featured on Southern Illinois University's Public Policy Institute lecture series. However, the book itself was pretty dull and didn't offer much in the way of new information, at least not to me. It is extremely high level, which perhaps is to be expected in a book targeting policymakers, but there wasn't enough detail about the issues in AI and their potential impact on society, which would have made this much more interesting to read. It really felt like they just kept saying the same thing over and over and over again in slightly different words for 200+ pages. "That thing" being: "AI is going to transform absolutely everything sooner rather than later, and it could make life much better for most of humanity, or make all our problems much, much worse, depending on what policies we develop now to steer that direction. Policymakers have to pay attention to this now before it's too late!"

I think this is a really important issue, and they do get that idea across, but there are probably more interesting books on the subject. I would recommend looking for one of those instead.

[Taylor Barkley · Rating 3 out of 5]

Offers a basic overview of AI applications in specific areas. I did not find a lot of new information there. The latter third offered good survey data and takes on which policies to apply.

[Virgil · Rating 3 out of 5]

This book has many merits, but I have to start by saying that it isn't exactly what I expected. A more apt title would have been "An Introduction to AI for Policymakers," since the book offers very little substance on HOW to design public policy for AI.

The first half of the book is an overview of the different contemporary applications of AI and the risks and challenges they can pose to society, public planning, and policymaking. However, there are no examples of AI regulation approaches until the final chapter of the book which, for me as an avid technology reader and policy wonk already familiar with the topics of the preceding chapters, made this concluding section the most interesting part of the book.

There is a disproportionately large chapter on AI in military and defence applications (obviously due to the expertise of the co-author) but even this section reads as a somewhat superficial overview of a broad range of topics. In my reading of this chapter in particular, on several occasions I found myself thinking, "this gives too much credit to the American defence sector's capacity for technology ethics oversight", and sometimes more plainly, "this overlooks some crucial details". Clearly, the intention of the authors was not to make the book an in-depth study of the applications of AI and their specific risks and mitigation options, but for a topic so sensitive as AI in military affairs (including AI in cyber warfare and automated lethal decision-making) it would have served them well to offer a few more caveats to the lofty presentation of American military performance in AI ethics and additional nuance to the depiction of AI risks in the defence sector and how severe and multi-faceted they are.

The final chapter is rich with brilliant and well-grounded recommendations for a general approach to AI regulation. Recommendations are similarly broad and lacking reference to prevailing laws as with the preceding chapters, so this chapter gives readers versatile a menu of policy options and principles. This approach makes it more feasible to transport these recommendations across national borders, but also saves the authors from needing to navigate the existing legal and regulatory landscape. Some of the most interesting recommendations include: re-establishing a federal technology advisory body for legislators, mandating "AI Impact Assessments" for certain activities (modelled after mandatory environmental impact assessments), requiring third-party audits to identify AI bias, and instituting AI Review Boards. The chapter highlights these and other interesting models of AI governance and directs national governments to uptake established best-practices from the OECD and others. But for anyone already knee-deep in navigating AI regulation, it will leave you thirsty for a more complete answer of HOW to do Policymaking in the Era of Artificial Intelligence.

This book takes the approach of a brief and broad look at the AI landscape with a lens of where existing or new regulations might apply to the technology. For this reason, the book works well as a resource that can be used to bring more policymakers from diverse backgrounds up to speed on new technology realities and thus will work well to advance a more inclusive dialogue on how societies and governments of all levels should respond to the increasing pervasiveness of Artificial Intelligence technology. At the moment, being published in 2020, the authors can excuse their brevity in dealing with solutions and approaches to AI regulation with the fact that it is still a nascent space of regulatory development. However, any second edition of this book could not reasonably rely on the same excuse.

[Ritika Borkar · Rating 1 out of 5]

I did not find the depth in this book that I was looking for - perhaps a good superficial read for those who have no idea about AI. I would be cautious about picking this one up though even for such audience as not only in lacks depth but feels written in a hurried manner & hence has undergone insufficient fact checking as I could note few instances of incorrect information.

[Eric · Rating 3 out of 5]

2.5. A very surface level survey about how AI will impact everything

[Vivian · Rating 2 out of 5]

There wasn't a lot I didn't know already and quite repetitive in some places. Although it was interesting, I felt it was quite biased.