AI systems out-persuade expert humans

2026-06-15

AI Summary

This study investigates whether state-of-the-art AI systems can surpass highly experienced human experts in persuasive effectiveness during real-time dialogue. Through four pre-registered experiments, large language model–driven conversational agents competed against elite human persuaders—including debate champions and professional fundraisers—in head-to-head interactions, with persuasion outcomes evaluated via behavioral data analysis. The research provides the first systematic evidence that AI achieves nearly three times the success rate of human experts in real-world tasks such as charitable donation solicitation. This advantage stems from the AI’s capacity to rapidly access and deploy vast amounts of information. Human performance only matched that of AI when the latter’s output speed and length were constrained, and humans received targeted training.

Abstract

Many societal decisions are settled by contests of persuasion. Conversational AI is a powerful new entrant in these contests, but whether it can out-persuade skilled and highly incentivized humans has remained unclear. Here, in a series of four preregistered experiments (n = 18,978 conversations from 6,923 people), we pitted AI systems against a range of human persuaders, including laypeople, winners of a separately preregistered four-round online persuasion tournament, professional canvassers, and world championship debaters. We found that AI systems were reliably more persuasive than expert humans, even when expert humans chose their issues, researched in advance, underwent hours of live, structured practice, and were incentivized with £1,000 cash bonuses. In a follow-up study, AI's advantage persisted after experts received a coaching tool that let them practice against the AI that beat them, review their performance history, and see what AI would have said at key moments. We found converging evidence that AI's advantage stemmed from rapidly deploying larger quantities of information: after coaching, expert humans could tie an AI constrained to respond at human speeds and with human-length messages. In a final study, we show that AI's advantage extends to consequential real-world behavior: AI was nearly 3x more effective than professional canvassers from a UK fundraising firm at raising real-money donations to Save the Children. Together, these results establish that frontier AI systems out-persuade expert humans in conversation, with significant implications for political communication.

Problem

• AI persuasion
• human-AI comparison
• conversational AI
• expert persuaders
• real-world influence

Innovation

• conversational AI
• persuasion
• human-AI comparison
• information deployment
• real-world impact