Autonomous Weapon Systems – Dangers and Need for an International Prohibition

Abstract

Advances in ICT, robotics and sensors bring autonomous weapon systems (AWS) within reach. Shooting without control by a human operator has military advantages, but also disadvantages – human understanding of the situation and control of events would suffer. Beyond this, compliance with the law of armed conflict is in question. Would it be ethical to allow a machine to take a human life? The increased pace of battle may overburden human understanding and decision making and lead to uncontrolled escalation. An international campaign as well as IT, robotics and AI professionals and enterprises are calling for an international ban of AWS. States have discussed about limitations in the UN context, but no consensus has evolved so far. Germany has argued for a ban of fully autonomous weapons, but has not joined the countries proposing an AWS ban, and is using a problematic definition. An international ban could comprise a prohibition of AWS and a requirement that each use of force must be under meaningful human control (with very few exceptions). If remotely controlled uninhabited weapon systems remain allowed, a-priori verification that they cannot attack under computer control is virtually impossible. Compliance could be proved after the fact by secure records of all communication and sensor data and the actions of the human operator. The AI and robotics communities could make significant contributions in teaching and by engaging the public and decision makers. Specific research projects could be directed, e.g., at dual use, proliferation risks and scenarios of interaction between two fleets of AWS. Because of high military, political and economic interests in AWS, a ban needs support by an alert public as well as the AI and robotics communities.