AI's Role in Modern Polling
Imagine an AI asking you about politicians. This is already happening. A French company called Naratis uses AI agents to conduct in-depth interviews. These agents check if you're answering correctly, if you need to elaborate, and if you're a real person.
Pierre Fontaine, Naratis's founder, says they are the first to use this for political polling. Other startups in the US, like Outset, Listen Labs, and Hey Marvin, use similar AI for commercial polling. This shift is making opinion research more automated.
Naratis focuses on qualitative research. This type of polling usually involves small groups or one-on-one interviews. These can take weeks and cost a lot. Naratis replaces this with AI conversations. They don't just ask people to pick answers. They explore how people form their opinions and when those opinions change.
We're a new kind of news feed.
Regular news is designed to drain you. We're a non-profit built to restore you. Every story we publish is scored for impact, progress, and hope.
Start Your News DetoxThe company claims its AI method is ten times faster and cheaper. It is also 90% as accurate as human polling. A study that once took weeks can now be done in a day or two. This speed comes from "parallelization," where many AI agents interview people at the same time.
Challenges and Future of AI Polling
AI polling comes at a time when survey response rates are very low. Stéphane Le Brun, an AI consultant, notes that rates have dropped from over 30% in the 1990s to below 5% today. This makes polling more expensive and less accurate.
Some critics point to past polling failures, like not predicting Brexit or Donald Trump's 2016 victory. Fontaine argues these issues mainly affect quantitative polling. Qualitative research, he says, is more about understanding opinions than predicting election results.
Established polling firms like Ipsos also use AI. They use it in market research to analyze videos of people's habits. This helps them observe behavior directly.
AI is also used to analyze social media. Researchers are experimenting with "digital twins" and "synthetic people." Digital twins are virtual models that respond like real individuals. Synthetic data creates new profiles based on real-world patterns. These tools help study small or hard-to-reach groups.
However, there are concerns. Ipsos does not use AI-generated respondents in political surveys. Bruno Jeanbart, CEO of OpinionWay, says they would never publish a poll based on AI-generated data due to trust issues.
AI systems can sometimes "hallucinate," giving incorrect answers. They might also produce "common sense" responses instead of actual opinions. If data is generated, what is truly being measured?
Trust is a big issue. Jeanbart believes countries like France might ban polls based on synthetic data. Experts agree that human oversight is still vital.
The future will likely be a mix of human and AI methods. AI will help with large-scale conversational surveys and faster insights. Digital twins and synthetic data might be used in market research.
In political polling, the line between using AI to help humans and using AI to simulate data will remain important. Companies like Naratis aim to change how opinions are gathered, turning surveys into conversations at a huge scale. The success of this shift depends on how the technology is used, explained, and regulated. Economic pressures will continue to drive more automation in the industry.
