Using AI as a Mock Interviewer

What you can practise

The hard part of preparing alone is the experience of being asked questions. Tell an AI the role and your background, and it will produce endless likely questions and follow-ups to your answers. Unlike a human partner, it does not tire of asking the same thing again. It cannot imitate a real interviewer's expressions, the atmosphere, or how each organisation actually assesses candidates, so use it to get comfortable with questions rather than to predict the outcome.

Set the scene and take one question at a time

Get twenty questions at once and you tend to read the list and stop. Set it up like a real interview: one question, your answer, then the next. Give the role and your experience level, but leave out your name and any internal details from your current employer.

• Example: "I want to practise for an interview for a marketing role with three years' experience. Act as the interviewer and ask one question at a time. When I answer, move on to the next one."
• Example: "Make the first five questions about my experience and motivation, then ask about a failure and a conflict at work."

Use follow-ups to find the gaps

In real interviews, people are thrown less by the first question than by the follow-up. Questions that dig into your answer, such as "so what was the result?" or "was there another way?", expose its weak spots. Asking for follow-ups on each sentence of your cover letter is another good check.

• Example: "Based on my answer, ask two follow-up questions an interviewer might use to dig deeper. Aim at anything I glossed over without evidence."

Set the feedback criteria first

Ask "was that good?" and you will mostly get praise and vague advice. Say what to judge against and the feedback becomes useful. Remember that the AI's verdict has no connection to whether you will actually be hired; trust it only for structure and clarity.

• Example: "Judge my last answer on three things only: is the conclusion in the first sentence, is the evidence a concrete experience, and could I say it in under a minute?"
• Does the conclusion come first?
• Is the evidence a specific moment from real experience?
• Did it actually answer the question, or drift elsewhere?

Why memorising model answers backfires

Asking the AI to write model answers and learning them by heart is not recommended. They are smooth but could come from anyone, so they leave no impression, and memorised sentences sound stilted. When a follow-up goes beyond what you memorised, the answer falls apart. Build answers from what you actually said, and remember the order of points and the key example rather than exact sentences.

Check company details yourself

Ask for "questions this company is likely to ask" and the AI may weave in facts about the company that are outdated or belong to another firm. Check the business and recent news at the original sources, such as the company website, the job advert and official filings, and base anything you mention at interview on those.

Out loud, against the clock

Answering in writing and answering aloud are different skills. Say your answers out loud and time them. Once an answer runs past a minute, the conclusion has usually slipped to the end. Record yourself on your phone and listen back to hear filler words and habits. A voice conversation feature is fine to use if available, but check where recordings are stored first.

After each session

Noting the following after each round makes the next one quicker.

• Questions and follow-ups that stumped you
• One example from experience for each answer
• Answers that ran over a minute and what to cut
• Company facts you checked yourself, with sources
• Notes on the order of points, not memorised sentences