Legal reviews of weapons, means and methods of warfare involving artificial intelligence: 16 elements to consider

Abstract

What are some of the chief concerns in contemporary debates around legal reviews of weapons, means or methods of warfare involving techniques or tools related to artificial intelligence (AI)? One session of the December 2018 workshop on AI at the frontiers of international law concerning armed conflict focused on this topic. In this post, I outline a few key threshold considerations and briefly enumerate 16 elements that States might consider as part of their legal reviews involving AI-related techniques or tools.

It is imperative, in general, for States to adopt robust verification, testing and monitoring regimes as part of the process to determine and impose limitations and—as warranted—prohibitions in respect of an employment of weapons, means or methods of warfare. Where AI-related techniques or tools are—or might be—involved, the design and implementation of legal review regimes might pose particular kinds and degrees of challenges as well as opportunities. With respect to challenges, for example, in a forthcoming blog post Netta Goussac will highlight several legal and other concerns that might arise in respect of reviews of weapons involving AI, not least the potential to introduce uncertainty and corresponding issues regarding (un)predictably and (un)reliability. Furthermore, today it seems, from my perspective, that sufficient trust among States in this area seems to be lacking, at least among certain States with advanced technological capabilities. Against that background, robust legal reviews may not only contribute to legal compliance, but may also help foster normative stability and augment trust among States.