A Virtue Ethics Approach to AI-induced Risk

Abstract

Risk plays an increasingly pivotal role in the regulation of digital technologies. At the EU level, for example, the General Data Protection Regulation, the Artificial Intelligence (AI) Act, and the Digital Services Act all embrace a risk-based approach. However, this paper contends that, across its various stages, risk management is a value-laden process, which extends beyond the legal and technical dimensions of risk. The integration of ethics into risk management is thereby necessary for the latter’s comprehensiveness and accuracy. To achieve such an integration specifically within the realm of AI, the paper begins by introducing the philosophical concept of risk. It examines AI-induced risk, in particular, from the dominant perspectives of consequentialism and deontology, highlighting their limitations in conditions of uncertainty. It moves on to alternatively suggest virtue ethics as an overlooked yet fruitful perspective focusing on the risk-taker’s character dispositions and contextual awareness. A preliminary virtue-ethical framework is thus advanced for the evaluation of AI-induced risk, intended for operationalisation by AI developers and regulators. The paper concludes by outlining the implications of this virtue ethics approach for research, policy, and practice.