Reviewing National Security Law for the Coming AI Revolution

REVIEW — We live in the age of Dreadnoughts. No, not of battleships, but of astounding leaps of technology, most particularly, Artificial Intelligence or AI. The Centaur’s Dilemma: National Security Law for the Coming AI Revolution by James E. Baker is at once a valuable tool for national security lawyers and policy makers and a critical lens into the ultimate question of whether AI’s legacy will be defined by increased volatility or reduced unpredictability.

Baker’s goal is to make AI and the associated legal issues accessible to national security lawyers and non-legal practitioners to promote “wise and strategic decisions about regulating the security uses of AI.” The book achieves its goal—and diverges from existing literature—by fashioning a compressive, accessible roadmap of the principles shaping society’s legal and policy response to AI innovations.

By avoiding the tendency of other AI essays to fixate on a single law or slender set of policy issues related to a particular aspect of AI, Baker’s unique contribution proves particularly useful to national security lawyers who wrestle with issues driven by emerging technology–Technology National Security Lawyers (TNSLs, pronounced “tensile” in our lexicon). The book title refers to the creature from Greek mythology that had the upper torso of a human and the lower body and legs of a horse, and represented the struggle between civilization and barbarism. The “Centaur model” as described by the Department of Defense refers to the placement of a human “in’ or “on” the loop for decision-making involving lethal autonomous weapons systems (LAWS). Baker translates this model into an effective framework, and reaches beyond the science fiction-induced fascination with LAWS to apply the framework broadly to the real-world advances that are profoundly influencing society’s approach to privacy and national security today.

The Centaur’s dilemma, as perceived by Baker, is how to reap the benefit of AI for broad national security purposes without losing control of the consequences. In Baker’s view, this dilemma is at the heart of decision-making in our national security arena.

Baker does not claim to give the definitive definition of AI, or offer a universal solution to other hardy perennial issues. His more valuable contribution is a useful legal framework of US domestic and international law and provides examples of historical analysis for study. The book is divided into two parts. The first portion describes AI, its security uses, and risks. The second part examines how, if at all, we should regulate the national security uses of AI through a close study of comparable policy problems and well-selected legal and historical case studies.

The book is divided into discrete sections and intended to be referenced “as needed, from the index,”—but policy mavens and TNSL lawyers alike would benefit immensely from Baker’s global understanding of the AI landscape—an epiphany bestowed to the reader who experiences this book cover to cover.

Baker applies constitutional analysis to AI issues and then elaborates on three domestic laws central to AI development in the United States: the International Emergency Economic Powers Act (IEEPA), the Invention Secrecy Act (ISA), and the Defense Production Act (DPA). His previous articles and analysis of the DPA relative to COVID 19 may be especially helpful to legal practitioners. Similarly, his discussion of various arms control regimes for international restrictions upon the development, production, proliferation and use of various weapons (Nonproliferation Treaty, Biological Weapons Convention, and Chemical Weapons Convention for example) merits closer analysis by lawyers, diplomats, historians, and national strategy professionals looking for analogs to the imposition of such limitations on AI via consensual treaties and agreements or by a nation or group of nations to enforce limitations upon a non-consenting country. The Convention on Certain Conventional Weapons (CCW) (formally known as the Convention on Prohibitions or Restrictions on the Use of Certain Conventional Weapons Which May Be Deemed to Be Excessively Injurious or to Have Indiscriminate Effects) and the ongoing dialog in the Group of Governmental Experts (GGE) since 2017 is briefly reviewed.

Baker also uses the Ottawa Convention (the Landmine Treaty) as a historical example, counterpoint, and warning to US government policymakers about diplomatic inaction on an issue where domestic and world popular opinion moved beyond the US government positions. A review of CCW and US experiences in past protocol negotiations, especially Protocol V, Explosive Remnants of War, and the more recent cluster munitions conventions, should be closely examined.

As the US and the GGE move forward on the mandate to examine emerging LAWS technologies, with a view toward developing recommendations for the clarification, consideration, and development of the normative and operational framework related to such weapon systems, our previous historic policy experiences should be lessons. CCW’s mix of military, technical, and diplomatic experts with involved NGOs could play a central role in US diplomatic efforts related to AI.

Baker’s discussion of the Outer Space Treaty provides another area of productive comparative study. Indeed, historical analysis of the space race, and especially superpower behavior over the decades related to negotiations on anti-satellite weapons are helpful analogs. AI and the race for AI dominance bear historical and strategic resemblance to dual-use space capabilities and the challenge of putting the genie back into the bottle. These analogs may also provide insight into negotiating from strength, unilateral versus mutual restraint, and linkages to other policy areas by competitors.

Baker reviews the use of AI for a host of national security applications, and the book examines possible futuristic uses of general AI for LAWS. But his disciplined gaze captures the more pressing and immediate concerns surrounding narrow AI and what our prime competitors in this area see as a Revolution in Military Affairs: AI-enabled command decision-making. China has witnessed its Sputnik moment in the Alpha Go series of contests and has invested in its version of the Manhattan Project to move decision-making from “informaticized” to “intelligentized.” China’s stated goal is to advance decision-making via AI so that human decision-making can no longer keep pace with the speed of decision-making in combat (i.e., the defeat of our Observe, Orient, Decide, Act (OODA) Loop).

This focus on AI driven decision-making seeks to leverage existing narrow AI technologies to ultimately achieve breakthroughs in general AI that will establish battlefield “singularity.” To that end, the recently released Center for Strategic and International Studies report, Maintaining the Intelligence Edge, makes several critical and long overdue recommendations: refocus the national intelligence priorities framework (NIPF) by raising the priority of science and technology (S&T) targeting and collection, elevate foreign S&T analysis to a core analytic discipline, and establish a technology net assessment function focused on emerging and disruptive technologies. Our failure to prioritize this targeting and collection places us in an informational and operational deficit with competitors. Baker’s book will empower TNSLS and operators to develop the norms and policies to drive an effective, agile, and moral national response to strategic competition in the AI space.

The competency of attorneys or national security lawyers expounding on AI and associated legal challenges is generally cause for some skepticism. Luckily, Baker has steeped himself in the subject matter, greatly benefiting (I suspect) from consultations with leading technical experts. The Centaur’s Dilemma should join Kai Fu Lee’s, AI Superpowers: China, Silicon Valley, and The New World, and the series of reports on AI from the Center for Security and Emerging Technology as required technology reading for the national security professional. Baker’s book excels as both a general reference and field guide. Most importantly, it is a clarion call heralding the immediate need to close a gaping law and policy void with intelligent governance and regulation. America’s lawyers and policy makers would be wise to heed the call.

Purchase this book

Book Review: The Centaur’s Dilemma: National Security Law for the Coming AI Revolution by James E. Baker / Brookings Institution Press

Reviewed by W. Renn Gade, General Counsel, Defense Intelligence Agency

The Centaur’s Dilemma earns a prestigious four out of four trench coats.

Disclaimer: The Cipher Brief, like other Amazon Affiliate partners, gets paid a small commission based on purchases made via the links provided in this review

Read more Under/Cover book reviews in The Cipher Brief

Read Under/Cover interviews with authors and publishers in The Cipher Brief

Interested in submitting a book review? Check out our guidelines here

Sign up for our free Undercover newsletter to make sure you stay on top of all of the new releases and expert reviews

Read more expert national security perspectives and analysis in The Cipher Brief