Smart questions, smart tech

A recurring feature on AI in health professions education

How do we teach future clinicians to use AI without eroding the clinical reasoning we’re working to build?

Artificial intelligence is no longer on the horizon for our learners. It’s in the exam room, the lab, the clinic, and the study session. Students across our health professions programs will graduate into a health care landscape where AI-assisted tools are part of daily practice.

That reality raises the question: If students lean on AI throughout their training, do they ever develop the diagnostic instincts, clinical judgment, and critical thinking that AI is supposed to augment — not replace?

Medical education researchers have given this challenge a few different names: “deskilling” (losing abilities we once had), “never-skilling” (failing to develop them in the first place), and “mis-skilling” (learning the wrong lessons from flawed tools). The Association of American Medical Colleges has made these risks a centerpiece of its national faculty development efforts, and the emerging consensus is encouraging. The answer isn’t to ban or surrender to AI. It’s intentional educational design — protecting space for active problem-solving early in training, while teaching students to use AI thoughtfully, transparently, and in service of the patient in front of them.

The UW School of Medicine and Public Health (SMPH) and our peer institutions are already engaging. At SMPH, we have an AI tool under development to assist faculty in giving pedagogically sound Objective Structured Clinical Examination (OSCE) feedback to students. At NYU Grossman, residents use an AI tool that records patient conversations and offers feedback on communication skills — AI as a coach for human connection. At Johns Hopkins, students work through AI-generated clinical cases, then defend their diagnostic decisions in dialogue with the tool. At the University of Virginia, students in a foundational clinical course learn to use AI alongside traditional resources to build differential diagnoses and treatment plans.

These approaches share a common thread that resonates with how we do things at SMPH: technology in service of people. The goal is to help shape better-prepared clinicians, stronger reasoning, and ultimately, better care.

Want to dig deeper?

• AAMC Webinar Series: AI Skill Building for Medical Educators — A free, recorded series covering deskilling risks, prompting strategies, curriculum design, and assessment.
• AAMC: Artificial Intelligence in Academic Medicine Webinar Series — Recorded sessions including “Preparing for AI Integration in Clinical Education” and insights from the Josiah Macy Jr. Foundation report on AI’s future in medical education.
• AAMC: Principles for the Responsible Use of AI in and for Medical Education (Version 2.0, 2025) — National guiding principles for integrating AI while safeguarding learning and professionalism.
• Deskilling Dilemma: Brain Over Automation — A concise commentary on protecting clinical reasoning and adaptive expertise in the age of AI.
• National Academies: Artificial Intelligence in Health Professions Education — Workshop proceedings that look beyond the MD degree to AI’s implications across the health professions, including physical therapy and public health.

Your turn

What AI questions are you asking in your program? Submit your AI-related questions to CurrentAffairs, and we’ll explore them in upcoming editions.