Fairness in agreement with European values: An interdisciplinary perspective on AI regulation

Abstract

" With the rise of digitalization, Artificial Intelligence (AI) is becoming ubiquitous. AI-based systems are being proposed and implemented to identify, optimize, automate, and scale solutions to complex economic and societal problems. These developments have spurred regulatory efforts, including the Proposal for an EU AI Act. This interdisciplinary position paper examines various concerns related to fairness and discrimination in AI and discusses how AI regulations address these issues, with a focus on the aforementioned Proposal. It examines AI and fairness through the perspectives of law, the AI industry, sociotechnology, and moral philosophy, presenting diverse viewpoints. Subsequently, it maps these perspectives along three axes of interest: (i) Standardization vs. Localization, (ii) Utilitarianism vs. Egalitarianism, and (iii) Consequential vs. Deontological ethics, revealing a pattern of common arguments and tensions among these axes. By situating the discussion within these axes and emphasizing the reconciliation of key tensions, the paper identifies and proposes the roles that AI Regulation should adopt to ensure the success of the AI Act in addressing AI fairness concerns."