Early studies suggest reliance on AI tools is eroding professionals' skills
Early studies reported by Nature suggest that leaning heavily on artificial-intelligence tools degrades the abilities of the people using them, with evidence emerging among both physicians and software engineers. The results are preliminary, but they point in a consistent direction: skills that go unpracticed because an AI system handles them tend to atrophy.
In medicine, the worry centers on clinicians who lean on AI detection and diagnostic aids and may perform less well when the assistance is removed. Similar concerns are surfacing in software development, where developers who delegate routine coding to AI can lose fluency in tasks they no longer do themselves.
The mechanism is unsurprising — capabilities erode without use — but the consequence matters because these are exactly the professionals expected to catch the tools’ mistakes. If the people meant to supervise AI become less able to do the work unaided, the safety margin that justifies deploying these systems narrows. Researchers caution that the evidence is still early and that tool design and training could mitigate the effect.