The Essential Role of Human Oversight in the Era of AI-Enabled Medical Assessment

Abstract

Artificial intelligence is rapidly entering health professions education, particularly in assessment where efficiency, scalability, and standardization are highly valued. While AI systems can generate items, score written responses, and support structured evaluations, their limitations become apparent when assessment involves complex professional behaviours, contextual judgment, and ethical responsibility. Drawing on Miller’s Pyramid, ethical frameworks, and emerging literature on AI-enabled evaluation, this editorial argues that AI may support assessment at lower cognitive levels, but cannot replace human oversight at the “Shows How” and “Does” levels where communication, professionalism, empathy, and moral reasoning are demonstrated. Without careful governance, AI risks reinforcing bias, misclassifying competence, and disadvantaging learners from culturally diverse or Global South contexts. The responsible path forward is not to reject AI, but to embed it within a human-in-the-loop model in which educators retain authority and accountability for high-stakes decisions. As the field embraces AI, the central question becomes not what AI can do, but where its authority must end.