Artificial intelligence is rapidly transforming the world of political opinion polling, replacing time-consuming human-led interviews with automated conversational systems capable of analysing public sentiment at scale.
"When you hear the word 'politician', what is the first image or emotion that comes to mind?"
The question is asked not by a human researcher, but by an AI-powered voice assistant. While a respondent shares his views over the phone, multiple AI systems simultaneously analyse the conversation. One verifies whether the person is answering the question correctly, another evaluates the depth of the response, while a third checks for possible fraud or bot-like behaviour.
The technology is being developed by Naratis, a French start-up focused on bringing artificial intelligence into political opinion research.
"The US has start-ups like Outset, Listen Labs and Hey Marvin that do AI polling like this in the commercial sphere. To my knowledge we're the first to do this for political opinion polling as well," says Pierre Fontaine, the 28-year-old engineer who founded the firm in 2025.
The emergence of AI-led polling marks a major shift for an industry traditionally dependent on manual interviews and extensive human analysis. In countries such as France, polling firms are increasingly exploring automation to reduce costs and speed up research processes.
Naratis specifically targets qualitative research, which is widely regarded as the most expensive and labour-intensive form of polling. Traditionally, these studies involve one-on-one interviews or focus groups that can take weeks to organise and analyse. By using conversational AI, the company says it can significantly reduce both time and cost.
Rather than relying on standard multiple-choice surveys, the platform encourages participants to engage in conversations with AI systems. "We don't ask people to tick boxes - they have a conversation with an AI," Fontaine explains. "That means we can explore not just what people think, but how they think - how they build their opinions, and even when those opinions change."
The company claims its approach is "10 times faster, 10 times cheaper and 90% as accurate as human polling".
According to the firm, projects that previously required weeks and substantial budgets can now be completed within a couple of days, with some responses collected in less than 24 hours. Fontaine describes this advantage as "parallelisation", where numerous AI agents conduct interviews simultaneously instead of relying on individual human researchers.
The rise of AI polling comes at a challenging time for the polling industry overall. Survey participation rates have dropped sharply over the decades, increasing operational costs and raising concerns about the reliability and representativeness of public opinion studies.
Supporters of AI polling argue that conversational systems may encourage respondents to be more honest, especially when discussing politically sensitive issues. Some researchers believe this could reduce social desirability bias, where people avoid expressing controversial opinions to human interviewers.
However, critics remain cautious about the growing dependence on AI in political research. Concerns include the possibility of AI systems generating inaccurate conclusions, producing overly generic responses, or creating misleading synthetic data.
Questions have also emerged around the use of "digital twins" and "synthetic people" — AI-generated profiles designed to imitate real human behaviour. While some market research firms use such tools for testing and simulations, many organisations remain reluctant to apply them in political polling.
At Ipsos, AI is already used extensively in consumer and behavioural research, including analysing user-recorded videos and studying social media activity. However, major firms continue to maintain human oversight in politically sensitive projects.
At OpinionWay, AI may assist with conducting interviews, but "we would never publish an opinion poll based on AI-generated data," says CEO of OpinionWay Bruno Jeanbart, citing concerns about trust.
Experts believe the future of polling will likely involve a hybrid approach combining AI efficiency with human supervision. While automation can accelerate research and lower costs, human researchers are still considered essential for validating findings, interpreting nuance and ensuring accountability.
Even AI advocates acknowledge the need for caution. "The goal is end-to-end automation, but today it would be unsafe and socially unacceptable to remove humans entirely," says Le Brun.
As economic pressures continue to push the polling industry toward faster and cheaper methods, companies like Naratis are betting that AI-driven conversations could redefine how public opinion is collected and understood. Whether this transformation strengthens trust in polling or deepens public scepticism may ultimately depend on how responsibly the technology is implemented and regulated.