Problem representation in the age of artificial intelligence: the state of a dying art?

Since its introduction into clinical reasoning education, the concept of the problem representation (PR) - a clinician's evolving mental model of a patient's illness - has helped frame how we teach reasoning to trainees. Despite PR's popularity and promise as an educational tool, its use has also prompted persistent questions about terminology, teaching methods, and its role in developing diagnostic expertise. The recent emergence of generative artificial intelligence (GAI) capable of rapidly synthesizing and summarizing clinical has raised new concerns: If GAI can perform PR, what clinical reasoning skills remain essential for learners to master? In this perspective, we explore the history, current state of the art, and projected future of PR in a world with widespread GAI use. We argue that problem representation remains a critical cognitive process for trainees to master on their path toward expertise, even as generative AI may change or improve how it is taught.