Techno-legal tinkering in war AI decision-support systems and international humanitarian law

Contrary to the popular promise that AI will improve and accelerate military decision-making, integrating AI into military operations is slow, complex, and fraught with challenges. It is also experimental in nature. The concept of “techno-legal tinkering,” introduced in this book, describes an ongoing process of modifying, retraining, and fine-tuning AI models that simultaneously shapes legal norms. The practice of tinkering allows the promises of AI's gradual improvement to obscure its inherent shortcomings. While aiming to rectify issues on an ad hoc basis, tinkering with AI decision-support systems concurrently introduces and normalizes uncertainties associated with their use.
“Techno-Legal Tinkering in War: AI Decision-Support Systems and International Humanitarian Law” offers a nuanced and critical examination of how AI reshapes the boundaries of permissible military conduct. Drawing on interviews with military practitioners, it explores attempts to achieve transparent battlefields through a “single pane of glass,” to encode the legal notion of reasonableness, and to increase users’ trust in AI output through confidence scores. In doing so, it traces how techno-legal interactions accompanying AI integration assemble (in)visibilities, (un)certainties, and (non-)knowledge in contemporary military contexts.