What AI Must Learn about Lethal Force

Autonomous weapons systems (AWS) raise among the most fundamental legal and ethical challenges of our times. Among them: How selecting persons to be killed by autonomous machine affronts human dignity, and second, how AI learning is unpredictable—it is a black box—meaning indiscriminate killing cannot be ruled out. The first issue requires a robust rule on meaningful human control; the second requires learning program data that includes only authentic law on preserving the peace and protecting human rights.