Three Pathways to Secure Greater Respect for International Law concerning War Algorithms

Abstract

Existing and emerging applications of artificial intelligence in armed conflicts and other systems reliant upon war algorithms and data span diverse areas. Natural persons may increasingly depend upon these technologies in decisions and activities related to killing combatants, destroying enemy installations, detaining adversaries, protecting civilians, undertaking missions at sea, conferring legal advice, and configuring logistics. In intergovernmental debates on autonomous weapons, a normative impasse appears to have emerged. Some countries assert that existing law suffices, while several others call for new rules. Meanwhile, the vast majority of efforts by States to address relevant systems focus by and large on weapons, means, and methods of warfare. Partly as a result, the broad spectrum of other far-reaching applications is rarely brought into view. One normatively grounded way to help identify and address relevant issues is to elaborate pathways that States, international organizations, non-state parties to armed conflict, and others may pursue to help secure greater respect for international law. In this commentary, I elaborate on three such pathways: forming and publicly expressing positions on key legal issues, taking measures relative to their own conduct, and taking steps relative to the behavior of others. None of these pathways is sufficient in itself, and there are no doubt many others that ought to be pursued. But each of the identified tracks is arguably necessary to ensure that international law is — or becomes — fit for purpose. By forming and publicly expressing positions on relevant legal issues, international actors may help clarify existing legal parameters, pinpoint salient enduring and emerging issues, and detect areas of convergence and divergence. Elaborating legal views may also help foster greater trust among current and potential adversaries. To be sure, in recent years, States have already fashioned hundreds of statements on autonomous weapons. Yet positions on other application areas are much more difficult to find. Further, forming and publicly expressing views on legal issues that span thematic and functional areas arguably may help States and others overcome the current normative stalemate on autonomous weapons. Doing so may also help identify — and allocate due attention and resources to — additional salient thematic and functional areas. Therefore, I raise a handful of cross-domain issues for consideration. These issues touch on things like exercising human agency, reposing legally mandated evaluative decisions in natural persons, and committing to engage only in scrutable conduct. International actors may also take measures relative to their own conduct. To help illustrate this pathway, I outline several such existing measures. In doing so, I invite readers to inventory and peruse these types of steps in order to assess whether the nature or character of increasingly complex socio-technical systems reliant upon war algorithms and data may warrant revitalized commitments or adjustments to existing measures — or, perhaps, development of new ones. I outline things like enacting legislation necessary to prosecute alleged perpetrators of grave breaches, making legal advisers available to the armed forces, and taking steps to prevent abuses of the emblem. Finally, international actors may take measures relative to the conduct of others. To help illustrate this pathway, I outline some of the existing steps that other States, international organizations, and non-state parties may take to help secure respect for the law by those undertaking the conduct. These measures may include things like addressing matters of legal compliance by exerting diplomatic pressure, resorting to penal sanctions to repress violations, conditioning or refusing arms transfers, and monitoring the fate of transferred detainees. Concerning military partnerships in particular, I highlight steps such as conditioning joint operations on a partner’s compliance with the law, planning operations jointly in order to prevent violations, and opting out of specific operations if there is an expectation that the operations would violate applicable law. Some themes and commitments cut across these three pathways. Arguably, respect for the law turns in no small part on whether natural persons can and will foresee, understand, administer, and trace the components, behaviors, and effects of relevant systems. It may be advisable, moreover, to institute ongoing cross-disciplinary education and training as well as the provision of sufficient technical facilities for all relevant actors, from commanders to legal advisers to prosecutors to judges. Further, it may be prudent to establish ongoing monitoring of others’ technical capabilities. Finally, it may be warranted for relevant international actors to pledge to engage, and to call upon others to engage, only in armed-conflict-related conduct that is sufficiently attributable, discernable, and scrutable.