Residency programs are exploring how artificial intelligence (AI) tools can strengthen the human relationships at the center of CBME. AI can be used to enhance coaching skills or help residents engage with their Individualized Learning Plans (ILPs). At the same time, these tools raise legitimate questions about oversight, bias, equity, and skill maintenance.When considering how to bring AI into your program's CBME infrastructure, keep the following in mind:
- AI augments coaching and assessment, but it doesn't replace faculty judgment. Tools such as simulated coaching conversations or milestone-mapping assistants are most useful as rehearsal space or a first pass, with a faculty coach or CCC member always reviewing and owning the final interpretation.
- Before residents or faculty start using AI tools, programs should establish and disseminate clear policies on which tools are approved, how resident and patient data are protected, and how AI-assisted work should be disclosed.
- Just as faculty supervise residents’ clinical reasoning, they need a framework for supervising how residents use AI, including knowing when to trust it, when to verify it, and when to set it aside.
- Deskilling is a documented risk of AI use. Heavy reliance on AI assistance can measurably erode a learner's unassisted performance, which is a reason to use AI as a scaffold for growth, rather than a permanent substitute for skill.
- Faculty development is a prerequisite, not an afterthought. Faculty members need working knowledge of what these tools can and can't do before they can supervise residents' use of them or interpret AI-assisted output responsibly.
- Always review AI output before acting on it. AI-generated coaching prompts, ILP goals, or milestone summaries can contain errors, generic language, or subtly off-base suggestions. The user should read and validate anything AI produces before it shapes a real conversation, plan, or assessment.
