The usual objection to AI-moderated research goes like this: people won't open up to a machine. A human moderator builds rapport, reads the room and earns honesty. An AI interviewer can only ask questions.
We wanted to test that, so we asked the people on the other side of the table.
We interviewed 50 US adults who had taken part in a live, human-moderated study within the past two years. They talked about that session, then compared it with the AI-moderated interview they were doing at that moment.
TL;DR: 1) people are just as comfortable with an AI interviewer as with a human moderator, and 2) when honesty does differ between the two, it shifts toward AI by more than five to one, because there's no one there to judge.
Key takeaways
An AI interviewer is software that runs a research interview. It asks questions, listens to the answers and follows up on what each participant actually said, with no human moderator in the session. AI-moderated research is qualitative research where an AI interviewer runs the sessions in place of a human moderator.
92% were very or extremely comfortable sharing honest thoughts with an AI interviewer. The figure for their last human-moderated session was also 92%.
34% said they're more honest with AI. 6% said they're more honest with a human, and 60% said it's about the same.
60% said an AI interviewer feels less judgmental and safer to be candid with. This was the most common reason by a wide margin.
58% would choose an AI interviewer for political or religious topics. A human moderator's strongest showing was physical and mental health, at 32%.
The main trade-off was warmth: 26% found AI more impersonal and 22% felt it lacked empathy or nuance.
An AI interviewer doesn't cost you rapport, and for the topics people find hardest to talk about, it gets more honest answers.
What is AI-moderated research, and how is it different from a survey?
A survey gives everyone the same fixed questions. An AI interviewer holds a conversation. It asks an open question, listens to the answer, and decides what to ask next based on what that person said: it probes a vague answer, asks for an example, or follows up on something unexpected.
AI-moderated research means letting that AI interviewer run the whole session in place of a human moderator. Participants usually join by video or voice, on their own schedule. Because there's no moderator's calendar to work around, hundreds of interviews can run at once instead of eight a day.
This study compares the two experiences from the participant's side: being interviewed by a human moderator, and being interviewed by an AI. We wanted to know whether talking to an AI interviewer changes how comfortable people feel and what they're willing to say.
What we found
1. Comfort: an AI interviewer matches a human moderator
We asked participants how comfortable they felt sharing their honest thoughts in their most recent human-moderated session. Later in the interview, we asked the same thing about the AI interviewer they were talking to.
The results were almost identical. 92% were very or extremely comfortable in both modes, and nobody chose "not very" or "not at all comfortable" for either.
The reasons behind that comfort were different, though.
With the human moderator, comfort came from the person. 68% credited the moderator's warmth, empathy and rapport, and 52% mentioned a non-judgmental framing and tone.
With the AI interviewer, comfort came from the person not being there. 60% said it felt less judgmental and safer, 34% said the predictable structure and pacing helped, and 30% valued doing it on their own time and in their own space.
"It makes me feel more open not seeing someone's face [...]"
"I have control over when I'm making the recordings.”
2. Honesty: when it shifts, it shifts toward the AI interviewer
For researchers, honesty matters more than comfort.
Most participants (60%) said their honesty is about the same either way. Of those who said it differs, the split was lopsided: 34% were more honest with AI, and only 6% were more honest with a human. No one said they were much more honest with a human.
The most common reason was social pressure. 46% said they're more honest with AI specifically because it takes away judgment, and 28% described holding back with humans to manage how they came across.
"AI isn't going to judge me [...] with a human I have to watch my words."
Human moderators still have an advantage for some people. 10% said a human makes them feel truly heard, and 8% said humans are better at probing and clarifying.
"The fact that it feels like someone is actually listening makes a difference [...] humans react better to follow-up questions."
In a human-moderated interview, the risk is self-censorship. In an AI-moderated interview, the risk is a shallow follow-up. Only the second can be fixed with a better discussion guide.
3. Topic: AI-moderated research wins where answers are sensitive
We asked participants which moderator they'd choose for research on seven different topics.
Preferences split by how sensitive the topic is:
Sensitive and polarizing topics lean toward AI. 58% would choose an AI interviewer for political or religious topics, 48% for relationships, and 46% for personal finances. In open-ended answers, 64% said sensitive or private topics are better suited to AI.
Health is the closest call. It's the topic where a human moderator does best (32%), although AI still edges ahead at 40%. Participants wanted a person who could recognise distress and refer them to help.
For low-stakes topics, most people don't mind. 68% had no preference for shopping habits, 64% for product feedback and 58% for entertainment.
"With an AI, you can program it to be neutral."
"I feel better talking to an AI about [finances] because the judgment aspect would be lacking."
"For anything health-related, I would prefer a human moderator... they can flag and refer me to care.”
When to use an AI interviewer vs a human moderator
Use an AI interviewer when… | Use a human moderator when… |
|---|---|
The topic is sensitive, private or polarizing (money, politics, relationships, habits people are embarrassed about) | The topic is emotionally complex and participants may need support (mental health, grief, serious illness) |
You need candor more than rapport | You need a moderator to adapt as things happen, for example co-creation or a live prototype walkthrough |
You need scale: dozens or hundreds of interviews, several markets, results in days | You have a small number of high-stakes stakeholder interviews |
Participants are spread across time zones or hard to schedule | The session depends on group dynamics in the room |
For many teams, the best answer is both. Use AI-moderated research for breadth and honest initial answers, then have a human follow up with the participants whose answers matter most.
Where Listen Labs fits in
Listen Labs is an AI interviewer built for qualitative research at scale. It runs AI-moderated video and voice interviews that follow up on what each participant actually said. It then codes the open-ended answers into themes you can measure, and links every finding back to the clip where it came from.
An AI interviewer that follows up. It asks a follow-up when an answer is vague and stays neutral on sensitive topics, which this study shows is where AI performs best.
Qual at scale. You can run hundreds of in-depth interviews at the same time, in 100+ languages, with results in days rather than weeks.
Mixed methods in one study. Rating scales, multiple choice, ranking and AI-moderated open-ended questions all sit in the same interview, so you see each score next to the reason behind it.
Traceable insights. Every chart and theme links back to the participant's own words.
Quality recruiting. You have access to 30M+ verified respondents with fraud screening, or you can bring your own customers.
Conclusion
The case for human moderation has always rested on rapport. This study suggests rapport mattered most because human moderators had to work against the pressure their own presence creates. Without a person in the room, comfort stayed the same, and the honesty that did shift moved toward the AI interviewer.
That doesn’t make human moderators obsolete. Participants still wanted a person for health topics, where they might need care or a referral, and some felt more heard by a human. But on politics, money and relationships, more of them chose the AI interviewer: 58% to 14% on politics and religion, and 46% to 18% on personal finances.
The usual objection to AI-moderated research is that people won’t open up to a machine. For the topics researchers find hardest to ask about, this study found the opposite: many participants would rather talk to the machine.
Methodology
Who: 50 completed interviews with US adults, recruited through Prolific. Participants had to have taken part in a live, moderated study (a focus group, one-to-one interview, ethnography or usability session) within the past two years. We analysed 55 interviews in total, and most figures are based on the 50 completes.
How: Each participant took one AI-moderated video interview on the Listen Labs platform. They described their most recent human-moderated session, rated their comfort, then rated and explained their comfort with the AI interviewer. Next they compared honesty between the two modes and chose a preferred moderator for seven topics.
Analysis: Closed questions are reported as percentages. Open-ended answers were coded into themes after the interviews. Because one answer can contain several themes, multi-coded charts add up to more than 100%.
Limitations: The study itself was AI-moderated, so read it as participant self-report, not a controlled experiment. Comfort with the human session was recalled, while comfort with the AI was reported in the moment. 94% of participants had done an AI-moderated interview before, so this sample is more familiar with the format than the general public.