New Verasight Study Shows: ‘You Shouldn’t Use AI to Ask What People Think about AI’

Randomized study finds AI-moderated interviews deliver significantly richer responses while raising concerns about participation and representativeness

SAN FRANCISCO, CA, UNITED STATES, September 2, 2026 /EINPresswire.com/ — Verasight, a nonpartisan survey research firm that collects verified, representative survey responses, today released the findings of its latest study showing that AI-moderated interviews can generate significantly deeper responses than traditional surveys, but at a steep cost to participation and representativeness.

In a randomized experiment of 3,160 U.S. adults, AI-moderated interviews generated 4.8 times more respondent words than traditional written open-ended questions. However, just 40.5% of those assigned to an AI interview completed it, compared with 99.4% of those answering written questions, and the attrition was not random.

AI-interview completers were more likely to regularly use AI, trust AI companies with their privacy and provide longer responses. As a result, completers were 7.4 percentage points more optimistic about AI than the full sample assigned to the AI condition. The AI interview group also showed demographic and political differences that remained even after weighting.

Using an AI interviewer to measure attitudes toward AI selects on the outcome. Agreement that AI will improve daily life runs 12.5 percentage points higher among AI-interview completers than among non-completers. That selection leaves the completer pool 7.4 points more AI-optimistic than the full sample assigned to the AI condition. A researcher who read only the AI-interview transcripts would, for example, conclude the public is more AI-optimistic than it is.

The study, “Depth at a Cost: A Randomized Comparison of AI-Moderated Interviews and Written Survey Open-Ends,” comes as AI-moderated interviewing is rapidly gaining traction as a way for researchers to conduct qualitative research at greater speed and scale. Verasight conducted the experiment to test whether these tools can deliver on that promise and what researchers may sacrifice in return.

“AI-moderated interviews clearly have the ability to get people to say more and provide researchers with significantly greater depth, but our research shows that depth comes with important tradeoffs,” said Ben Leff, CEO and Co-Founder of Verasight. “When nearly six in 10 people assigned to an AI interview don’t complete it, and the people who remain are systematically different from those who leave, researchers need to be careful about treating those responses as representative of a broader population.”

Participants were randomly assigned to answer the same questions through either traditional written open-ends or an AI-moderated video or audio interview, where the AI could ask adaptive follow-up questions.

Those follow-up questions drove roughly 80% of AI interviewing’s advantage in response depth. AI interviews also generated 1.8 times as much usable information per assigned respondent, drawing out more specific examples, personal context and explanations for why respondents held certain views.

That added depth came with significant attrition. While 99.4% of participants assigned to written questions completed the survey, only 40.5% assigned to an AI interview did so, with most dropoff occurring when respondents were redirected to the external interview platform.

Despite the high dropoff, respondents who completed the AI interviews generally liked the experience. Among the completers, 86% rated it a four or five out of five, 91% said it was easy to complete and 34% said they believed they were more honest with a machine interviewer than they would have been in a standard survey.

The findings suggest AI-moderated interviews may be the most useful option when researchers prioritize depth, discovery and understanding why people hold certain views. For research requiring broad participation and representative population estimates, traditional written surveys or mixed-mode approaches remain more appropriate.

The full study, “Depth at a Cost: A Randomized Comparison of AI-Moderated Interviews and Written Survey Open-Ends,” is available at www.verasight.io/reports/ai-moderated-interviews-vs-survey-open-ends.

Survey Methodology
Verasight surveyed 3,160 participants from May 22–31, 2026, randomly assigning them to either traditional written open-ended questions or an AI-moderated video or audio interview. Both groups were asked about the same four topics, allowing researchers to compare completion, response depth and representativeness across the two formats.

About Verasight
Verasight is a nonpartisan survey research company founded to expand access to high-quality survey data. Built by researchers and methodologists, Verasight combines multi-mode recruitment, 100% in-house data collection, and continuous verification to ensure every response comes from a real person. Its infrastructure powers academic research, government studies, and national media coverage, including polling data for ABC News, The Associated Press, CBS News, CNN, FOX News Media, and NBC News.

Stephen Francy
Rubenstein PR
+1 212-805-3017
sfrancy@rubensteinpr.com
Visit us on social media:
LinkedIn

Media gallery