An Ethics Framework for Autonomous Weapon Systems

This essay explores the challenges faced by militaries and governments as they seek to develop, deploy and operate weapon systems that have increasingly autonomous elements, and goes on to propose an Ethics Framework for Autonomous Weapon Systems (AWS). The capabilities of artificial intelligence (AI) are expanding rapidly and concerns about how AI might be used, especially in AWS, have prompted serious reflection by governments and international organisations. However, even theorising around AWS is problematic and for several reasons across multiple fields: ‘autonomy’ remains a hotly contested term; AI takes multiple forms with different functionalities, and several types can be used in a single system; ‘predictability’ in AI is hard to anticipate but is crucial to satisfy both the ethics, legality and operational reliability of AWS; an understanding of the extent and nature of human-machine interactions within an autonomous weapon system is not yet mature; new ethical considerations are added to already-disputed ethics of war; future communications rely on nascent quantum entanglement capabilities; and human and machine bias is an ever-present risk. To begin to outline an ethics framework for AWS, this paper conceptualises a future lethal autonomous aircraft system (LAAS) – popularly referred to as lethal autonomous drones – by taking the Reaper MQ-9 remotely piloted aircraft system as a starting point and considering the implications of how a future variant might operate when elements of its functionality are delivered autonomously using AI. The paper concludes by outlining an ethics framework for AWS under four headings: overarching ethics of war, human responsibility, the methods and means of war, and risk assessment and mitigation.