Book Review: Autonomous Weapons Systems and International Norms

Technological advances in machine learning, predictive analytics, and machine vision have allowed for the development of weapons systems that incorporate artificial intelligence (AI) into the target selection and engagement process. While many states are readily adopting this technology, the enhanced autonomy associated with Autonomous Weapons Systems (AWS) threatens to erode existing international norms, including the norm of meaningful human control over the use of force. In Autonomous Weapons Systems and International Norms, Ingvild Bode and Hendrik Huelss argue that states’ practices and behaviour, rather than formal legal negotiations, have already played a critical role in establishing international norms and “standards of appropriateness” for AWS.

Proponents of AWS often cite the presumed military advantages associated with heightened degrees of weapons autonomy. Indeed, states’ interest in deploying these systems stems in part from their presumed ability to improve battlefield analysis, overcome enemy countermeasures, and reduce costs. ¹ However, AWS also pose considerable risks, including the erosion of international norms regarding the role of human decision-making in warfare. In light of this, scholars and activists have used international forums such as the Convention on Certain Conventional Weapons (CCW) to promote norms that would place legal restrictions on military autonomy and the use of AWS. However, Bode and Huelss point to disagreements between participants about the legal definition and acceptability of AWS and the intentional efforts on behalf of some states to promote vague or indeterminate rules to demonstrate that formal, deliberative negotiations are often unable to formulate clear, comprehensive standards of appropriateness. Importantly, the challenges associated with the development of formal norms are not unique to AWS; the authors note that there has historically been a “wide gap between the codification of rules and the practice of warfare,” whereby formal rules have often been vague, implemented as reactive responses to humanitarian disasters, or simply disregarded in times of war. ²

In light of the challenges associated with establishing norms through multilateral deliberations, Bode and Huelss seek to broaden the field’s understanding of the norm-construction process. While the authors maintain that norms can be deliberately negotiated and codified into international law, they introduce the concept of “procedural norms” to argue that practices and behaviours also play a central role in the establishment of standards of appropriate conduct. These norms stem from nonverbal, unintentional practices, and can inform, challenge, and erode formal legal norms, including those relating to the use of force. Using the concept of procedural norms, the authors demonstrate that standards of appropriateness relating to the use of AWS have already emerged, despite the relative novelty of AI-powered weapons. ³

In particular, the authors examine the ways in which the deployment of air defence systems that can select and engage targets autonomously has eroded the norm of “meaningful human control,” which maintains that humans must retain final decision-making powers over the use of force. ⁴ More specifically, they argue that ‘‘states using air defence systems have…implicitly accepted the compromised role of human agents,” concluding that “meaningless human control has become accepted by a group of states as an emerging ‘appropriate’ understanding of how force can be used.” ⁵ Thus, the practice of autonomous air defence has influenced standards regarding meaningful human control in ways that will inform norms surrounding AWS.

While Bode and Huelss make a convincing case for the role of practices in the norm-setting process, this book leaves the reader with a number of unanswered questions. First, the authors argue that procedural norms highlight the ways that actors beyond the state, including individuals, groups, and communities, influence international norms. However, the bulk of their analysis centres on the actions and behaviour of states, with little consideration for the role the public plays in norm construction. Given growing public interest in and anxiety about the weaponization of AI, it is likely that the development of AWS will generate considerable backlash from large segments of the population. Indeed, public opposition to the deployment of weapons that use AI to independently identify and strike targets will likely be much larger than opposition to defensive weapons systems that use automation, but not AI. It is unclear what role, if any, public opinion and anxiety about the weaponization of AI will play in state practice and the development of norms. This therefore leaves readers to wonder: whose idea of “appropriateness” matters when setting norms on the use of force?

Second, readers may question whether norms that have emerged from states’ practices with air defence systems necessarily will apply to AI weapons with offensive capabilities. The normative and ethical implications of autonomous air defence systems, designed to identify overhead threats and strike in response, may differ from those related to AI-powered systems that can identify and engage their targets with limited human involvement. While the air defence systems examined in this book have resulted in human casualties and certainly do inform perceptions of meaningful human control in an era of enhanced weapons autonomy, states and their citizens may feel differently about systems that were deliberately designed for offensive purposes. Thus, it remains unclear whether norms derived from practices with automated defence systems will necessarily apply to the AI-powered weapons systems of the future.

Finally, the authors’ claim that international norms regarding the use of AI-powered weapons have already emerged before the full extent of AWS has even been realized may strike some readers as fatalistic. If legalistic, formal norms are fundamentally insufficient and procedural norms have already begun emerging in ways that erode meaningful human control, what can be done to set restrictions on the development and proliferation of these weapons?

Ultimately, Bode and Huelss’ Autonomous Weapons Systems and International Norms makes a timely and important contribution to the fields of International Relations and International Security Studies at a time when global interest in AI and its military applications is growing. This book provides a comprehensive overview of the literature on international norms and employs rich, detailed case studies to analyze the evolution of norms surrounding AI-powered weapons, making it important reading for those interested in international norms and emerging technologies.