KeyLIME+ [38] AI and Emerging Skill Gaps: Deskilling, Never‑Skilling, Mis‑Skilling—and What Educators Can Do About It All
Mar 17, 2026
Raja-Elie Abdulnour, a pulmonary and critical care physician and medical educator at Brigham and Women’s/Harvard and NEJM editor, explores how AI can quietly erode clinical reasoning. He outlines risks like deskilling, never‑skilling, and mis‑skilling. He introduces the DEFT-AI framework and contrasts centaur versus cyborg approaches to using AI in training.
AI Snips
Chapters
Transcript
Episode notes
Pause And Verify AI Outputs
- Teach learners to pause and verify AI outputs instead of accepting them automatically.
- Use AI as an informant to reasoning only after asking learners to question its evidence and compare with their judgement.
AI May Widen Performance Gaps
- AI benefits are uneven: higher-performing clinicians gain more while lower-performing clinicians can be harmed.
- Studies (Harvard Business School and Nature Medicine) showed lower-competence users made more AI-related errors than experts.
Secure Core Skills Before Adding AI
- Ensure learners attain core competencies before introducing AI augmentation.
- Raja-Elie recommends strengthening baseline skills so trainees cross a competence threshold and benefit rather than be harmed by AI.