New Framework Keeps Medical Skills Sharp in an AI-Powered World
Future doctors could lose critical skills if AI is used the wrong way—this research shows the fix.
AI is now part of how doctors learn and practice. But educators worry that trainees are leaning on it too much and losing essential skills like diagnosis and clinical judgment. This paper from medical education researchers reframes the issue: it's not about how much AI you use, but whether you choose the right "mode" of AI interaction for each task. They call this mistake "misclassification," not "misuse."
The team proposes a framework called SCAN—short for Substitute, Complement, Aid, and Non-Negotiable. Think of it as a simple menu for medical educators. For routine tasks, AI can substitute human effort. For complex tasks, it can complement or aid the learner. And for foundational skills, AI should be non-negotiable—kept out so the trainee practices without help. The goal is to keep AI as a teammate, not a crutch.
The paper also identifies a hidden danger: passive engagement. When a learner simply watches an AI perform a task correctly, they may feel like they're learning but actually are not. Over time, this quietly produces "mis-skilling"—doing things wrong without realizing it. The authors suggest that when this happens, the AI should step back and let a human expert take over, acting as a quality check.
This framework gives medical schools a concrete way to decide when to deploy AI in training, supervision, and testing. It also opens the door to more research on how to prevent skill erosion. For patients, this matters because better-trained doctors make better clinical decisions—and that directly affects the quality of care you receive.
- The SCAN framework sorts AI use into four roles: Substitute, Complement, Aid, and Non-Negotiable, giving educators clear rules for when AI should and shouldn't step in.
- The real danger is 'passive engagement'—watching AI work instead of doing the task yourself, which can silently erode clinical skills.
- When a learner isn't improving, the paper says AI assistance should be replaced by human expert guidance to keep training on track.
Why It Matters
Smart rules for when doctors use AI mean safer, more capable physicians—which directly protects patient health and trust.