Conversational AI has emerged as a formidable force in persuasion, outperforming even highly skilled and incentivized human experts, according to a new study published on arXiv. The research, conducted through a series of four preregistered experiments, pitted AI systems against a diverse set of human persuaders and found that the AI consistently won.
The Experiment and Participants
The study involved a total of 18,978 conversations from 6,923 people. The human persuaders included laypeople, winners of a separately preregistered four-round online persuasion tournament, professional canvassers, and even world championship debaters. In one condition, expert humans were allowed to choose their own issues, research in advance, undergo hours of live structured practice, and were incentivized with £1,000 cash bonuses. Despite these advantages, the AI systems were reliably more persuasive.
Key Findings
| Comparison | Result |
|---|---|
| AI vs. expert humans (best conditions for humans) | AI reliably more persuasive |
| AI after experts received coaching tool | AI's advantage persisted |
| AI constrained to human-speed responses vs. coached experts | Experts could tie the constrained AI |
| AI vs. professional canvassers (real-world fundraising) | AI nearly 3x more effective |
The coaching tool allowed human experts to practice against the AI that beat them, review their performance history, and see what the AI would have said at key moments. Only when the AI was constrained to respond at human speeds and with human-length messages could expert humans tie the AI. This suggests that AI's advantage stemmed from rapidly deploying larger quantities of information.
Real-World Impact
In a final study, the AI's persuasive power extended to consequential real-world behavior. The AI was nearly 3x more effective than professional canvassers from a UK fundraising firm at raising real-money donations to Save the Children. This demonstrates that the advantage is not limited to laboratory settings.
Implications for Technology Leaders
For CTOs and enterprise technology buyers, the findings underscore the potential of AI in any domain requiring persuasion — from sales and marketing to customer service and negotiation. The ability of AI to rapidly process and deploy information gives it an edge over human experts. However, the research also shows that human experts can close the gap when AI is limited to human communication speeds, indicating that the advantage is partly due to speed and volume of information. Organizations investing in conversational AI for business outcomes should consider both the raw persuasive power and the ethical implications of deploying such systems at scale.
