Distribution of Forward-Looking Responsibility in the EU Process on AI Regulation

Abstract

Artificial Intelligence (AI) is beneficial in many respects, but also has harmful effects that constitute risks for individuals and society. Dealing with AI risks is a future-oriented endeavor that needs to be approached in a forward-looking way. Forward-looking responsibility is about who should do what to remedy or prevent harm. With the ongoing EU policy process on AI development as a point of departure, the purpose of this article is to discuss distribution of forward-looking responsibility for AI development with respect to what the obligations entail in terms of burdens or assets for the responsible agents and for the development of AI. The analysis builds on the documents produced in the course of the EU process, with a particular focus on the early role of the European Parliament, the work of the High-Level Expert Group on AI, and the Commission's proposal for a regulation of AI, and problematises effects of forward-looking responsibility for the agents who are attributed forward-looking responsibility and for the development of AI. Three issues were studied: ethics by design, Artificial General Intelligence (AGI), and competition. Overall, the analysis of the EU policy process on AI shows that competition is the primary value, and that the perspective is technical and focused on short-term concerns. As for ethics by design, the question of which values should be built into the technology and how this should be settled remained an issue after the distribution of responsibility to designers and other technical experts. AGI never really was an issue in this policy process, and it was gradually phased out. Competition within the EU process on AI is a norm that frames how responsibility is approached, and gives rise to potential value conflicts.