The Concept of Autonomy
[Law and the Future of War Research Paper No 2] This paper investigates the notion of autonomy as it applies to software and cyber-physical systems, with a focus on matters which bear some significance to the application of relevant bodies of international law. As autonomy, in the context of software, is a technical term rather than a legal or philosophical one, a bottom-up approach is taken, beginning with a description of the technical origins and meaning of the term ‘autonomous’. Based on that, two important relationships are discussed: that between the autonomous system and its environment, and between the system and its operator. Finally, several aspects of autonomy of relevance to a legal analysis are discussed: autonomous systems do not necessarily behave differently to manually operated systems; autonomous systems are not ‘independent’ of humans for the purposes of a legal analysis; and the relative contributions of human and machine in an operation involving an autonomous system are likely to be complex and variable, such that they should arguably be viewed as a form of collaboration between human and machine rather than a simple delegation of the entirety of a task to a machine.