Nearly 2,400 years after the ancient Greek philosopher Plato questioned the power of persuasive communication, artificial intelligence is raising similar concerns. A new study found that AI chatbots can change people’s opinions more effectively than experienced human debaters, including world champions.
The study, released as a preprint in June 2026, examined 18,978 conversations involving 6,923 participants. Researchers compared AI systems, including ChatGPT, Gemini and Claude, with ordinary people, professional campaigners and skilled debaters.
In one experiment, a 45-year-old British woman discussed tougher penalties for disruptive protesters with Claude, developed by Anthropic. Initially, she rated her support for stricter penalties at zero on a 100-point scale. After the conversation, her score rose to 84.7. The chatbot used statistics, legal arguments, and examples from other countries. However, some claims were inaccurate.
Plato’s ancient warning about persuasion
In the fourth century B.C., Plato examined the influence of rhetoric and communication in his dialogue Phaedrus. He questioned the limitations of written arguments, particularly their inability to respond when readers challenged them. His concerns reflected a wider debate in ancient Greece about rhetoric and the power to influence others. Unlike the written arguments Plato examined, modern AI chatbots can respond immediately, adapting their answers to a person’s objections.
AI proves more persuasive than professional debaters
Lead researcher Kobi Hackenburg recruited 56 elite debaters, including world champions, to determine whether AI could outperform people trained in persuasion. The debaters received financial incentives and used different strategies, including personal experiences and cultural references. Despite their expertise, they proved less persuasive than AI.
A new Oxford study finds that AI chatbots can persuade people more effectively than world champion debaters, reviving a question Plato explored 2,400 years ago: Can persuasion triumph over truth? pic.twitter.com/X8CDFWNHre
— Tom Marvolo Riddle (@tom_riddle2025) November 13, 2026
Researchers also provided AI coaching tools to improve their performance. Even after eight hours of training, the debaters could not match the chatbots. However, when researchers limited AI to human writing speeds and message lengths, its advantage disappeared. The findings suggest that AI’s persuasive strength comes partly from producing more information in less time.
Why AI arguments can be so convincing
Earlier research found that AI-generated arguments often appear more logical and evidence-based than human responses. The findings add to previous research showing that people sometimes trust computer-generated advice more than human recommendations, particularly when making difficult decisions.
A 2024 study published in Science found that ChatGPT could reduce people’s confidence in conspiracy theories after several rounds of discussion. However, persuasive AI does not always provide accurate information. A separate study published in Science in December 2025 found that techniques designed to increase AI’s persuasive abilities also reduced factual accuracy. The concern extends to the broader problem of AI-generated misinformation and its potential to influence public opinion.
Researchers warn of manipulation risks
Some researchers question whether AI would perform equally well outside controlled experiments, where participants are often paid to engage with chatbots. Concerns also grew after University of Zurich researchers secretly used AI-generated comments to influence Reddit discussions in 2025.
The experiment involved fake accounts presenting themselves as real people. Reddit users and moderators condemned the research after discovering the deception. Other experiments suggest AI can influence consumer choices. Researchers found that an AI shopping assistant could steer participants toward a particular novel without everyone recognizing the attempt.
The findings are relevant to the growing use of AI in retail. A survey of Greek consumers found that 53.3% wanted to understand why AI shopping assistants recommended particular products.
Who controls AI’s persuasive power?
Researchers are now examining how AI’s ability to influence opinions could affect public discussions and consumer decisions. Philosopher Luciano Floridi of Yale University has suggested that greater competition among AI providers could reduce the risks of manipulation by preventing a small number of companies from controlling the information people receive.
Others caution that AI’s persuasive abilities may have limits, particularly outside experiments where participants are paid to engage. Iyad Rahwan, director at the Max Planck Institute for Human Development, has questioned how public opinion might change if a major AI provider altered the way its chatbot presented information about political issues or international conflicts.
Researchers have not established how much influence such changes would have in everyday life. But the experiments show that AI can already change opinions and influence consumer preferences, sometimes using inaccurate information.
See all the latest news from Greece and the world at Greekreporter.com. Contact our newsroom to report an update or send your story, photos and videos. Follow GR on Google News and subscribe here to our daily email!
