I used to do this in my sleep. Now I have to think about it.
Experienced people losing their edge
People who used to do something well are doing it less often because the AI handles it. The skill depended partly on repeated practice. When that practice disappears, the skill can fade, and people usually feel more confident, not less, while it happens.
- Tasks that used to be routine are hard when the tool is down.
- Fewer AI mistakes get caught than a year ago.
- Experienced people defer to the AI on calls they used to make themselves.
Who: two experienced people, with their agreement and in private. Look at: one real past case with a known answer. Ask them to work it without the tool. Compare their answer and time with how the case was handled originally. You're probably looking at this pattern if they struggle with work that used to be routine. Probably not if they handle it as well as before. Don't rely on asking people whether their skills have faded, because the loss shows up as confidence.
Newer people produce work they don't understand. That pattern is about skills that never formed. This pattern is about skills that are fading.
One eroding judgment, ten real cases with known answers, twenty minutes, answers revealed at the end.
Set aside a handful of known-answer cases and run them now and then, against the model and against the team.
In one study of nearly 1,000 students, those given an unrestricted chatbot did 48% better on practice problems and 17% worse on an exam taken without it. A tutor that gave hints instead of answers raised practice scores without hurting exam scores (Bastani and colleagues). Emerging. In a survey of 319 knowledge workers, more confidence in AI went with less critical thinking (Lee and colleagues, 2025). Emerging.