Beyond Automation: A Conceptual Framework for Medical Responsibility and Human Competencies in AI-Driven Healthcare

Background/Objectives: Artificial intelligence (AI) is increasingly embedded in clinical reasoning, documentation, prediction, communication, and delegated digital and physical action via agentic and embodied systems. The central ethical challenge is no longer limited to identifying which tasks AI can automate, but determining which human competencies become essential as technological capabilities expand. Methods: This concept paper employs a five-step analysis: technological horizon mapping, selective synthesis of work-transformation evidence, conceptual analysis of medical action, normative synthesis, and medico-legal interpretation. Results: We propose a four-domain framework addressing four core questions of responsible medical action: epistemic (What is true?), normative (What ought to be done?), anthropological–relational (Who is the subject of care?), and systemic–hermeneutic (What does this mean here and now?). Grounded in a conception of technology as augmentative alteration, the framework recognises that technological capabilities can enhance beneficial action but also amplify harm; technological capability alone therefore cannot determine its appropriate use. We also distinguish interpersonal relationships from clinical human–machine interaction, propose professional exigibility as an analytical medico-legal category, and argue that meaningful human oversight requires more than mere human presence. Conclusions: The findings outline key implications for clinical practice, medical education, healthcare organizations, regulation, and a future empirical research agenda focused on human–AI error convergence.