Governing AI Technologies in Weapon Systems from the Bottom Up: Practices to Sustain and Strengthen Human Control

Autonomous and AI technologies have become integrated into the targeting functions of weapons.

Algorithmic rather than human decision-making may therefore prevail in warfare, raising humanitarian, legal, ethical, and security concerns. Currently, there are no specific, legal rules to govern autonomy and AI in the military. In their absence, ERC AutoNorms has found that use of autonomous and AI technologies in weapon systems makes norms.

Practices of design, of training personnel for, and of using such weapons shape a norm of what counts as the requisite form of human control over the use of force.

AutoNorms has revealed that this emerging norm accepts a diminished, reduced form of human control when designing and using military autonomous/AI technologies as “normal”.

This emerging norm is a societal challenge and a public policy problem because it undercuts the meaningful exercise of human agency over life and death decisions in war.

The insight that practices make norms holds the potential for social innovation as a process of recontextualising norms through enacting different practices.

Stakeholders could therefore shape a positive norm of human control through changing their practices from the bottom up.

The ERC PoC project AutoPractices initiates and accompanies this social innovation process in three ways via multi-stakeholder collaboration: first, AutoPractices raises awareness among stakeholders about the differential effect of their practices on the emerging human control norm in the military domain via empirical mapping.

Second, AutoPractices co-creates a set of best practices with stakeholders to sustain human control in weapon systems integrating autonomous and AI technologies at three workshops.

Third, AutoPractices diffuses the co-created set of best practices via an operational toolkit and dissemination through multipliers.

In this way, AutoPractices becomes a source of diffusing a positive norm of human control in weapon systems from the bottom-up.