Naratis uses conversational AI for political opinion polling

9 articles · Updated · BBC.com · Apr 30

Founded in France in 2025 by engineer Pierre Fontaine, the company says its method is 10 times faster, 10 times cheaper and 90% as accurate as human polling.
AI agents interview respondents simultaneously, probe answers and screen for fraud, as pollsters face response rates falling from above 30% in the 1990s to below 5% today.
Established firms including Ipsos and OpinionWay are also adopting AI, but remain wary of synthetic respondents in political surveys because of hallucination, trust and possible future regulation.

With AI polling promising speed and savings but facing accuracy and trust issues, what hybrid solutions could truly revolutionize opinion research?

Can AI-driven polling ever win public trust, or will concerns about authenticity and privacy always limit its adoption in sensitive areas?

How might regulatory and ethical standards evolve to address the risks of AI-generated data shaping public opinion and policy decisions?