Human-machine (learning) interactions War and law in the AI era

Technology, law, and war evolve together. This thesis explores this evolution in the contemporary era of artificial intelligence (AI). Today, Machine Learning (ML), a technique of AI, pervades many aspects of daily life, including in military operations. Whilst typically the legality of emergent military technologies is assessed under International Humanitarian Law (IHL), this study reframes the inquiry by asking how these technologies shape IHL targeting law assessments when military practitioners use ML-enabled systems to support decision making. Taking a practice-oriented approach, targeting law assessments are conceptualised as interpretative and rhetorical practices of legal reasoning. Human agency is situated, and legal meaning generated, within the everyday practices of IHL interpretation. The question is whether ML technologies will transform warfare and, in doing so, alter the nature of targeting law assessments.
To examine how the use of ML intervenes in battlefield legal assessments, an account of ‘human-machine (learning) interaction’ (HMLI) is advanced that puts the interaction between practitioners and ML technologies centre stage. While practitioners may retain the capacity to make targeting decisions when using decision-support systems, these technologies may nevertheless enable and constrain the choice architectures within which decisions are made. HMLI holds space both for how human agency is exercised in IHL targeting law assessments and the normative influence of technological affordances. Situating HMLI within practices of legal reasoning, this study explores how the use of ML can reconfigure targeting law assessments and generate new legal meaning in IHL interpretation along three lines: recontextualising factual understanding; introducing logics of quantification into qualitative IHL standards; and reshaping legal categories through statistical normalisation.