The one interview question nobody prepares for

"Walk me through how you actually use AI in your current role — and where you deliberately don't."

A version of this is in every mock interview we run, and it separates people faster than any coding round. Good answers are boring and specific: a named tool, a real workflow, a number, then a genuine limit. "I don't let it touch anything I can't check myself" beats an enthusiastic list every time.

Bad answers are all the same shape — a tour of everything they've tried, no judgement about where it breaks.

Try it in the replies. I'll give you honest feedback.