In a bold leap towards the future of opinion polling, the French start-up Naratis is harnessing the power of artificial intelligence to transform how public sentiments are collected and analysed. By employing AI agents in place of traditional interviewers, Naratis aims to enhance both the speed and depth of qualitative research, offering a fresh perspective on the complex world of politics.
A New Era of Polling
Imagine a scenario where an AI agent poses probing questions about political perceptions, with multiple AI systems working in tandem to ensure the responses are genuine and insightful. This is precisely what Naratis is doing, with founder Pierre Fontaine leading the charge since the company’s inception in 2025. The firm’s innovative approach allows for a qualitative polling process that is not only ten times faster and cheaper than conventional methods but also claims to achieve up to 90% accuracy compared to human-led polling.
Fontaine, a 28-year-old engineer, emphasises that the goal isn’t just to gather data but to understand how opinions are formed and evolve. “We don’t ask people to tick boxes; they have a conversation with an AI,” he explains, highlighting the depth of understanding that can be achieved through this method.
The Challenge of Declining Response Rates
The introduction of AI into polling comes at a critical moment, as response rates have plummeted from over 30% in the 1990s to below 5% today, according to industry expert Stéphane Le Brun. This decline poses a significant challenge for traditional polling methods, which rely heavily on human engagement. As fewer individuals are willing to respond, the accuracy and representativeness of surveys suffer, leading to an erosion of public trust.
Naratis’s AI-driven model seeks to counteract these trends by facilitating large-scale engagement through conversational interfaces. Unlike quantitative polling, which primarily counts responses, Naratis focuses on the richness of the conversation, aiming to explore the nuances of public opinion.
Navigating the Landscape of AI in Polling
While the advantages of AI in polling are clear—speed, cost-effectiveness, and the ability to gather richer data—concerns linger about the reliability and ethical implications of this technology. Critics point to notable polling failures, such as the unexpected outcomes of Brexit and Donald Trump’s presidential election, questioning the validity of AI-driven insights. However, Fontaine argues that qualitative research, which Naratis excels in, is less about predicting outcomes and more about understanding public sentiment.
Established polling firms are also adapting to this AI wave, incorporating technologies to analyse behavioural data more effectively. For example, Ipsos has begun using AI to interpret video footage of respondents, allowing for a more direct observation of behaviours rather than relying solely on self-reported answers.
Nevertheless, caution is exercised in politically sensitive polling. Companies like OpinionWay assert they will not publish AI-generated data due to trust issues, indicating a need for human involvement to validate findings.
The Future of Polling: A Hybrid Approach
As the landscape of polling continues to evolve, the most probable scenario is a hybrid model where AI and human expertise coexist. While AI can augment traditional polling methods by facilitating large-scale surveys and integrating social media insights, the role of human oversight remains crucial for ensuring the integrity and reliability of the data collected.
The introduction of digital twins and synthetic data may find niche applications in market research, but the political polling domain will likely maintain a strict boundary between human respondents and simulated data. Naratis’s innovative approach aims to shift the conversation around polling, turning mere surveys into engaging dialogues that yield unprecedented insights into public opinion.
Why it Matters
The implications of AI in polling extend far beyond mere convenience. As technology reshapes how opinions are gathered and understood, it could either restore public trust in the polling process or deepen existing scepticism. The future of political engagement may hinge on how well these technologies are integrated, explained, and regulated, making it essential for the industry to navigate this delicate balance with integrity and transparency. As economic pressures drive further automation, the responsibility lies in harnessing these advancements wisely, ensuring they serve to enhance, rather than undermine, democratic discourse.
