Human oversight over high-risk artificial intelligence systems. Contributions from International Humanitarian Law and European Union law

Abstract

The automation of decision-making by artificial intelligence systems is a growing phenomenon affecting all areas of society. The European Commission, aware of the risks that the use of these technologies entails for fundamental rights and freedoms, proposes in its Artificial Intelligence Act to introduce human oversight as a mandatory requirement for the design and development of these technologies. However, human oversight is underdeveloped in the European regulatory environment. For this reason, we propose to resort to the concept of Meaningful Human Control developed in the framework of International Humanitarian Law. In this article we analyse the state contributions, doctrinal and policy proposals made by the Group of Governmental Experts on Lethal Autonomous Weapons Systems. All these contributions allow us to approach from a novel perspective the concept of human oversight proposed by the European Commission for high-risk artificial intelligence systems. We conclude the article looking for universally applicable elements in human oversight for the automation of decision-making, irrespective of the field in question.