Regulation of AI: Learnings from Medical Education

AbstractThe rapid integration of artificial intelligence (AI) into health care presents regulatory challenges due to the dynamic, opaque, and adaptive nature of AI - particularly with generative AI models. The U.S. Food and Drug Administration has proposed a life-cycle approach extending beyond conventional frameworks, recognizing that traditional medical device regulations are ill-suited for such technologies. This perspective draws parallels between AI regulation and competency-based medical education (CBME), an existing framework that addresses similar complexities in training human clinicians who are also dynamic general-purpose problem-solvers with opaque cognitive processes. We propose adopting an AI-CBME life-cycle framework for medical AI regulation, leveraging the five core CBME components: defining competencies, sequenced progression, tailored learning experiences, competency-focused instruction, and programmatic assessment. By applying these principles, we can implement continuous outcomes-based assessments of AI systems within their operational environments. This approach embeds real-time validation and safeguards for patient safety, building accountability and trust, while maximizing the potential of AI technologies to enhance health care. Engaging medical educators in this process offers an immediate and practical pathway to develop a robust regulatory framework aligned with existing educational methodologies.