Empowering medical students with AI literacy: A curriculum development journey

The integration of artificial intelligence (AI) in healthcare necessitates AI literacy within medical education. As AI's role in health care expands, understanding algorithm transparency, accountability and bias is crucial. However, incorporating AI education into an already dense curriculum poses challenges. A structured, efficient course covering both technical and ethical aspects of AI is essential to prepare future clinicians for AI-enabled health care.

We developed a one-credit, 18-hour AI literacy course for medical students, balancing theoretical foundations with experiential learning. The course structure comprised a 3-hour lecture on fundamental AI concepts, two 6-hour hands-on workshops where students worked in groups of three to four and a concluding 3-hour discussion and reflection session. These sessions were strategically designed to ensure engagement while accommodating students' demanding schedules. Shorter, more frequent sessions were considered but deemed impractical due to scheduling constraints and the challenge of effectively conducting hands-on activities in a fragmented format.

The course was initially introduced in 2020 and 2021 for second-year medical students, attracting 11 and 13 students, respectively (23% of the cohort). Based on student feedback, it was revised in 2022 to target senior students (fifth- and sixth-year), increasing participation to 33%. In the workshops, students developed and deployed AI models (e.g., knee fracture detection, wound segmentation), guided by a data scientist and a clinician with expertise in the AI topic, fostering interdisciplinary collaboration.

Key topics like privacy, bias, data security and patient autonomy were integrated into projects, prompting reflection on social impacts such as ethical AI use and healthcare disparities. Project themes were selected based on faculty expertise and contemporary AI applications, ensuring clinical relevance. Student learning was assessed using a 17-competency framework,¹ measuring AI literacy before and after the course to evaluate effectiveness and inform future improvements.

Transitioning the course to senior medical students enhanced engagement and comprehension, aligning AI concepts with clinical applications. Quantitative assessments showed substantial improvements in AI literacy, particularly in ‘AI's strengths and weaknesses’ (RS 1.6), ‘data literacy’ (RS 1.3), ‘critically interpreting data’ (RS 1.15) and ‘ethics’ (RS 1.15). Constructive feedback from students, collected via structured surveys, highlighted the value of hands-on experience, interdisciplinary learning and real-world AI applications.

The design and implementation of our 18-hour AI literacy course provide insights into integrating AI education within medical training. First, while AI education programmes vary in length—from brief workshops to full-semester courses—our approach demonstrates that an intensive yet feasible structure enables medical students to develop key competencies within a compact timeframe. Second, curricula should emphasize hands-on learning, guiding students through real-world AI challenges and ethical considerations. Third, AI literacy training may best target senior medical students with more clinical experience, preparing them to use AI independently. Lastly, challenges included diverse student technical backgrounds and the rapid evolution of AI, requiring continuous faculty upskilling. These challenges highlight the need for adaptive AI curricula that evolve with technological advancements and learner needs.

Ming-Yuan Huang: Conceptualization; writing—review and editing; writing—original draft; methodology.