A competency milestones framework for AI-augmented clinical training and practice

What was the educational challenge? AI technologies are entering clinical workflows faster than training programs can implement competency frameworks that can ensure AI is used safely and responsibly. Concerns regarding automation bias, de-skilling, and never-skilling are increasingly discussed across specialties, but developmental frameworks that address these risks within clinical training and practice remain limited.What was the solution, and how was it implemented? We developed a competency milestones framework for AI-augmented clinical training and practice through iterative analysis of AI-assisted clinical workflows in GI endoscopy, breast radiologic imaging, and outpatient internal medicine documentation, cross-referenced with an AI competency milestones framework for health professions educators. The resulting framework comprises five domains: (1) AI Foundations and Clinical Data Literacy; (2) Ethical AI Use; (3) Safety, Governance, and Systems Integration; (4) Independent Clinical Performance and Skill Preservation; and (5) Patient Communication, Transparency, and Professionalism. Each domain has five developmental levels from Novice to Expert, with Level 2 serving as a readiness gate before independent AI-augmented practice.What lessons were learned, and what are the next steps? This framework can make it explicit when clinicians are ready for AI-augmented practice and how they should continue preserving independent clinical judgment and skill. Next steps include feedback from broader groups of clinicians and health professions educators and pilot implementation in specialty-specific scenarios.