JMIR Publications

ABSTRACT

Artificial intelligence (AI) is rapidly reshaping clinical education by embedding assessment and feedback within everyday learning activities. Medical students can now access AI platforms at any time, run complex clinical scenarios, and receive individualized feedback within seconds. Recent reviews show that AI is increasingly integrated across medical curricula, supporting simulation, diagnostic reasoning, and knowledge acquisition. Yet the presence of more data and more feedback does not automatically translate into better learning. The central challenge for medical educators is not whether AI can generate feedback, but how AI generated information can be curated and used as genuine assessment for learning. In this Viewpoint, we define artificial intelligence assisted formative assessment as the intentional use of AI tools to provide personalized feedback that explicitly serves learning rather than grading. We describe how AI supported systems can reveal detailed patterns in learner performance and expand opportunities for safe deliberate practice, while also introducing risks such as automation bias, overreliance on algorithmic scores, and neglect of relational aspects of care. We argue that the educational value of AI assisted formative assessment depends on how human educators interpret and apply AI outputs. We then outline three practical domains for action in medical education: supporting learner agency in working with AI, designing AI assisted formative activities in specific curriculum contexts, and preparing faculty and institutions to use these tools responsibly. Rather than proposing new technology, this Viewpoint offers a framework for clinical educators to transform AI generated feedback into a small number of actionable changes that strengthen student agency, uphold professional values, and sustain assessment for learning in the AI era.