Make Your AI Interviewer Argue Back: Persona Design for More Honest Answers

For decades, the way we ask research questions has been shaped by one hard constraint: a person is asking another person. Rapport, neutrality and careful, non-leading phrasing all exist because the interviewer is human, and because the person answering reads the face, the tone and everything left unsaid. AI-moderated interviews quietly remove that constraint. An AI moderator is unoffendable and endlessly consistent, and it can hold a persona no human interviewer could sustain, or would ever be allowed to attempt.

That opens up styles of questioning that were simply not available before. This session tests the most interesting of them: deliberately making the interviewer provocative and argumentative, so it pushes back, challenges what people claim, and even imposes an assumption on them to provoke a correction. When someone is nudged into defending what they believe, they tend to tell you far more, and far more honestly, than when they are asked politely and neutrally.

Tim Treagus runs the same interview through three deliberately engineered personas, a neutral moderator, a warm and encouraging one, and a provocative one that argues back, across several markets. The session shows what each persona pulls out of people, where provocation earns better data, where it backfires, and how to build these personas into your own prompts, with the guardrails that keep challenge constructive rather than hostile.

This is not a tool demo. It is a more technical and more psychological way of thinking about how we ask questions once the interviewer is no longer human.

Main practical takeaways

1. Treat interviewer persona as a design decision. Right now most AI interviews inherit whatever default tone the model happens to have. You will leave able to choose that tone on purpose and match it to your research objective.
2. Prompt patterns for an interviewer that pushes back. Concrete, reusable ways to make an AI moderator challenge a claim, surface a contradiction, or float an assumption for the person to correct.
3. A read on what persona actually changes. How warmth and provocation move depth, candor and the gap between what people say and what they do, so you can predict the trade-off before you field.
4. When to provoke and when not to. Provocation carries a drop-off and rapport risk. You will get a clear sense of the topics and moments where it pays, and where a neutral or warm persona serves you better.
5. Guardrails and ethics. How to keep a challenging interviewer constructive, protect participants, and stay on the right side of the line before you put this in front of real people.
6. A wider point about what is now askable. Questions and question styles that were never possible or permissible human to human are now on the table, and that is worth designing for deliberately.