Autonomous Weapons and the Laws of Targeting: Mapping Use-Cases to Identify Operational Constraints

Abstract

This paper aims at studying constraints posed on autonomous weapon systems (AWS) by their tasks and environments. We find use-dependent functions need stronger standardisation, while design-related ones require significant technical safeguards. We then introduce a categorisation of the weapons’ use-case. We find the added-value of AWS rests in defensive missions, either for access denial or last-resort self-defence. Other scenarios, excepted strategic strikes, present significant drawbacks. We conclude a reactive model of autonomy seems a more realistic realisation of AWS than the fully autonomous system most feared by some. This is significant because it may comply with requirements of meaningful human control and military operational control.

Keywords: autonomous weapons, humanitarian law, meaningful human control, humanity, distinction, proportionality