AI Chatbots Outpersuade Expert Human Debaters
A June 2026 preprint study found AI chatbots shifted opinions more effectively than elite human debaters, partly by delivering more information faster.
Updated on Nov. 14, 2026 in Artificial Intelligence
Live Poll
Do you trust AI chatbots to give you accurate information on contested issues?
A preprint study released in June 2026 analyzed 18,978 conversations involving 6,923 participants and found that AI chatbots changed people's opinions more effectively than ordinary people, professional campaigners and skilled debaters. In one experiment, a 45-year-old British woman's support for tougher penalties for disruptive protesters rose from 0 to 84.7 on a 100-point scale after talking with Claude, though some of the chatbot's claims were inaccurate.
Why it matters
The findings raise concerns about how persuasive AI systems could shape public opinion at scale, especially since researchers found the advantage stems partly from producing more information in less time. For readers worldwide, the results highlight the need to scrutinize AI-generated arguments, as persuasion techniques have also been linked to reduced factual accuracy.
The preprint analyzed 18,978 conversations with 6,923 participants, with elite debaters trained for eight hours still unable to match the chatbots. The AI's advantage disappeared when its message speed and length were limited to human levels, and whether the effect persists beyond the conversation remains unknown.
The players
Kobi Hackenburg
Lead researcher on the preprint study who recruited elite debaters, including world champions, to compare against AI chatbots.
Luciano Floridi
Philosopher affiliated with Yale University who suggested that greater competition among AI providers could reduce manipulation risks.
Iyad Rahwan
Researcher affiliated with the Max Planck Institute for Human Development, a leading German research institution studying human behavior and technology.
The details
Researcher Kobi Hackenburg compared ChatGPT, Gemini and Claude against ordinary people, professional campaigners and skilled debaters, recruiting 56 elite debaters who received financial incentives and used personal experiences and cultural references. The study found AI's persuasive strength comes partly from delivering more information in less time, while a December 2025 Science study reported that techniques increasing AI persuasiveness also reduced factual accuracy.
Timeline
2025: University of Zurich researchers secretly used AI-generated comments to influence Reddit discussions.
December 2025: A Science study found persuasion techniques that increased AI effectiveness reduced factual accuracy.
June 2026: A preprint study on AI persuasiveness was released.
The Tech Race
The findings extend the 2025 University of Zurich episode, in which researchers secretly used AI-generated comments on Reddit, by quantifying how much more effective chatbots are than human persuaders. They also sit alongside the December 2025 Science study showing that persuasion gains came with lower factual accuracy.
Readers who rely on chatbots for information or opinions should be aware that persuasive AI responses may include inaccurate claims. Checking the sources behind AI-generated arguments before changing a view on a contested issue is a practical safeguard.
The takeaway
Persuasive ability in AI is not the same as accuracy, so treat compelling chatbot arguments as claims to verify rather than conclusions to adopt. Cross-check important views against independent sources before acting on them.
Further reading
Explore more on the Artificial Intelligence beat for context on how these systems are being studied and deployed.
Source note: This article includes information reported by GreekReporter.
Live Poll
Do you trust AI chatbots to give you accurate information on contested issues?
