Used

Applied for a senior role I'd been targeting for a while. Used AI heavily to prepare — system design, behavioral questions, the works. The answers felt polished and solid going in. Bombed the interview. Not just underperformed — bombed. The company actually sent written feedback afterward, which almost never happens. Three of my answers contained factual errors. Not vague or shallow — factually wrong. One involved a distributed systems concept I'd apparently been misunderstanding for years, and the AI just gave me a more articulate version of my own misconception. That's the part that stuck with me: the AI doesn't know what you don't know. It can't flag the gaps in your mental model. It reflects your existing understanding back with more fluency — which means confident errors become even more confidently expressed. I'd essentially used AI to make myself sound more certain about things I didn't fully understand. What's actually helped since: using AI to generate hard questions, answering them completely without AI first, then checking my answer. The AI as examiner, not coach. Much more uncomfortable. Much more useful. Anyone else run into this? Or found a prep workflow that actually stress-tests your understanding instead of just polishing how you present it?