In a groundbreaking shift, the landscape of opinion polling is being redefined as artificial intelligence steps into the spotlight. The French-based company Naratis is pioneering this change, harnessing AI to conduct qualitative political polling at unprecedented speed and efficiency. Founded in 2025 by engineer Pierre Fontaine, Naratis aims to tackle the traditional pitfalls of polling, ultimately reshaping how public sentiment is measured and interpreted.
The Dawn of AI Polling
Imagine engaging in a conversation with an AI, discussing your views on politicians without the constraints of tick-box surveys. This is exactly what Naratis offers, as its innovative AI agents engage respondents in meaningful dialogues. While traditional polling methods often rely on time-consuming interviews and small focus groups, Naratis replaces this laborious process with advanced conversational AI technology.
Fontaine explains that this approach allows researchers to delve deeper into the nuances of public opinion: “We don’t ask people to tick boxes—they have a conversation with an AI. This means we can explore not just what people think, but how they think.” The result? A method that is reportedly ten times faster, ten times cheaper, and 90% as accurate as conventional human polling techniques.
Addressing the Challenges of Traditional Polling
The rise of AI polling comes at a time when traditional methods are facing significant challenges. Response rates have plummeted from over 30% in the 1990s to less than 5% today, leading to concerns over the representativeness and reliability of polling data. As fewer people respond, polling becomes increasingly costly and, inevitably, less trusted by the public.
Yet, Fontaine is optimistic. He argues that the shortcomings of past polling failures—such as the unexpected outcomes of Brexit and Donald Trump’s election—primarily impact quantitative polls. Qualitative research, which Naratis focuses on, is less about predicting election results and more about understanding the intricate tapestry of public opinion.
The Role of AI in Modern Research
Naratis is not the only player in the field; established firms are also integrating AI capabilities into their polling methodologies. For example, Ipsos employs AI to analyse video footage of participants discussing their habits, enabling deeper insights into behaviour. Meanwhile, AI is being utilised to evaluate social media trends and create digital twins—virtual models designed to mimic real individuals for behavioural analysis.
Given the current climate of distrust in polling, AI-driven methods offer a promising solution by facilitating quicker, more flexible data collection. Additionally, the impersonal nature of AI can yield more candid responses, particularly on sensitive issues. This could explain why traditional polling in France has historically underestimated support for the far-right.
The Risks and Ethical Considerations
However, the advent of AI polling is not without its risks. Critics caution that AI systems can sometimes “hallucinate,” producing misleading or entirely fabricated responses. Moreover, the use of synthetic data—where responses are generated rather than collected from real people—raises significant ethical questions about the validity and interpretation of results.
Human oversight is crucial in this evolving landscape. While AI can enhance the efficiency of polling, it cannot fully replace the need for human judgement in validating results. Fontaine acknowledges this, stating that “the goal is end-to-end automation, but today it would be unsafe and socially unacceptable to remove humans entirely.”
Why it Matters
As the polling industry grapples with declining trust and participation rates, the integration of AI presents both an opportunity and a challenge. Naratis and its contemporaries are not simply looking to replace traditional methods; they’re striving to transform the way public opinion is understood. The potential for faster, more insightful data collection could reinvigorate interest in polling, but this will heavily depend on the ethical frameworks and regulations established around AI usage. The future of opinion polling will hinge on finding the delicate balance between machine efficiency and human insight, ensuring that the voices of the public are not only heard but accurately represented.
