Abstract The rapid integration of Artificial Intelligence (AI) in critical sectors such as healthcare has raised significant concerns regarding human autonomy and the risk of delegating vital decisions to opaque artificial systems. This paper examines the normative foundations required for the responsible integration of ethical principles into AI-based Clinical Decision Support Systems. First, it reviews key descriptive models of human moral judgment—rationalist, intuitionist, and integrative approaches—and highlights their relevance for understanding how humans make ethical decisions under conditions of uncertainty. Particular attention is given to the Actor–Deed–Consequences (ADC) model, which captures the plural and intuitive structure of moral evaluation and offers a promising basis for informing ethical guidance functions in AI systems. Building on this analysis, the paper argues that descriptive models alone are insufficient to secure the legitimacy of AI-assisted clinical decisions. To address this gap, it proposes a discursive ethics framework grounded in the work of Karl-Otto Apel, Jürgen Habermas, and Adela Cortina. This framework emphasizes communicative rationality, stakeholder involvement, transparency, and accountability as essential conditions for ethically acceptable AI deployment. The paper concludes that AI-based CDSS should be developed as dialogical and process-oriented tools that support human judgment. Embedding discursive ethical procedures within AI architectures provides a robust normative pathway for ensuring that technological innovation in healthcare enhances human agency, safeguards dignity, and strengthens trust in clinical practice. Keywords: AI Ethics, Discursive Ethics, Clinical Decision Support Systems, Human Agency, AI Act, Communicative Rationality.
José-Félix Lozano (Thu,) studied this question.