JMIR Publications

ABSTRACT

Background:

Artificial intelligence (AI) is becoming increasingly relevant in clinical practice. Medical students are therefore expected to understand its principles, applications, and ethical implications. However, there lacks consensus on what AI content should be included and how it should be taught. Existing literature describes curricular content, teaching approaches, or local implementation initiatives but has not been synthesized to guide curriculum design. This lack of clarity limits informed decision-making regarding curricular standardization.

Objective:

This narrative review aims to elucidate the key AI content to be taught to medical students, as well as the pedagogical methods to implement such curricula.

Methods:

This narrative review synthesized published literature on AI as a subject of instruction in undergraduate medical education. We conducted a systematic search of two databases to identify the proposed curricular content or pedagogical strategies for teaching AI to medical students. Data was extracted and synthesized according to adapted PRISMA-ScR guidelines, allowing for comparison of content domains and teaching approaches across studies.

Results:

Twenty-seven studies were included. Most curricula featured multiple components. Foundational AI knowledge was almost universally included, although the depth of technical content varied widely. Ethical, legal, and data security considerations were also consistently prioritised. Clinical applications of AI featured prominently in various medical specialties, most commonly through workflow optimisation to contextualise learning. Pedagogically, didactic teaching was common but rarely used in isolation; structured hands-on practice and project-based learning were frequently incorporated to support the practical application of AI tools. Few studies explicitly referenced learning theories to inform curriculum design or pedagogical choices.

Conclusions:

Undergraduate AI education commonly integrates foundational concepts, ethics, and clinical application. Teaching approaches are often blended, but pedagogical choice is rarely explicit. Future curriculum development should define baseline AI competencies while offering differentiated learning pathways.