Positioning Artificial Intelligence Topics in Medical Education: A Curricular Relevance Perspective (Preprint)

UNSTRUCTURED Today we are witnessing a rapid introduction of AI applications into medical practice. However, their integration into medical education lags behind and remains uneven, often leaning more toward applied tools than fundamental conceptual understanding. In many curricula, students are exposed to AI instruments and applications without any detail regarding how such systems actually represent knowledge, reason, learn, or may even fail. From an educational perspective, this situation raises concerns both about the quality of the learning process and about the safe, aware use of AI in clinical context. Based on my experience developing and teaching an elective course on Artificial Intelligence in Medicine for medical students, I argue that a major challenge lies in justifying the inclusion of fundamental AI concepts for learners with limited theoretical background and minimal clinical exposure. Although these concepts are often perceived as overly abstract or difficult, their omission can undermine students’ ability to appropriately interpret AI generated outputs and to recognize system limitations. Building on this perspective, I propose the introduction of a Curricular Relevance Index (CRI) as a conceptual framework for curriculum design in medical AI education. The CRI characterizes AI topics across five dimensions: professional (clinical) relevance, foundational necessity, risk of omission, accessibility for the target learner population, and the intended depth of coverage. By explicitly separating foundational necessity from perceived clinical relevance, the CRI provides a transparent rationale for why certain “theoretical” topics remain essential, even when their immediate clinical visibility is limited. In addition to CRI, for practical considerations, an Instructional Practicality Framework (IPF) is proposed, which qualitatively captures pedagogical feasibility in terms of teaching methods, technical engagement, and resource requirements. Thus, balanced curriculum design decisions can be made that account for both educational value and implementation constraints. Therefore, this viewpoint offers a structured perspective on curriculum design, grounded in iterative teaching practice and student feedback. While illustrated in the context of early undergraduate medical education, the proposed frameworks are adaptable and can be transferred to other educational settings.