
Geoff Tomaino
People Don’t Want to Argue with AI, and That Makes Them More Persuaded by It
Our research began with a broad question: could AI’s biases eventually become our biases? For that to happen, people would first need to be receptive to AI’s opinions. That made it important to understand whether, and under what conditions, people are actually persuaded by AI models to change their minds.
Persuasion research has traditionally focused heavily on the message itself. Is the argument strong? Is the source credible? Is the message personalized? Does it use the right tone? These are important questions, and a growing literature has begun asking whether AI can generate persuasive messages as effectively as, or even more effectively than, humans.
But conversational AI introduces something different. A conversation is interactive. We do not simply receive information; we respond to it. The recipient of a persuasive message is also an active participant in the exchange.
That led us to think about AI persuasion in a different way. Rather than asking only whether AI produces better arguments than humans, we asked whether people themselves behave differently when they are conversing with AI rather than with another person.
In our studies, participants discussed a matter of opinion, for example, whether it is better to exercise in the morning or in the evening. Participants were told that they were speaking either with an AI chatbot or with another human participant. In reality, the replies in both conditions were generated by a large language model instructed to persuade participants toward the opposite of their initial position. This allowed us to isolate the effect of believing that one’s conversational partner was an AI rather than a human. Across these studies, participants argued less with the AI, and, by extension, were persuaded more by it.
We found that this tendency not to argue with AI was rooted in people’s lay beliefs about its persuadability. In pilot studies, participants explicitly rated AI as less persuadable. When we asked why, many described AI as incapable of genuinely holding an opinion, or as having a position that was effectively hard-coded and fixed. In other words, they did not approach AI as another mind whose position could meaningfully be changed through argument.
That belief changed how people responded when AI tried to persuade them. When another person disagrees with us, disagreement naturally invites a response: “No, that’s not right,” “But what about…,” or “Here’s why I think otherwise.” We are often motivated to defend our own position and, in turn, to change the other person’s mind. With AI, that impulse was weaker. Because participants saw little point in trying to persuade the AI, they argued back less—and, as a result, became more open to the opposing position their conversational partner was advancing.
This suggests that whether AI’s biases ultimately become our biases will not depend only on how persuasive these systems are at generating arguments. It will also depend on how people interact with them. In our case, AI became more persuasive in part because people engaged in less counterarguing against what it said.
That possibility is especially important as conversational AI becomes embedded in ordinary decisions about what to buy, where to travel, what to eat, what to watch, and how to organize daily life. Its influence may not come only from having better information or producing better arguments. It may also come from occupying a peculiar social position: something we are willing to listen to, but do not always feel is worth arguing with when we disagree.
Ongoing thread. More from Geoff Tomaino to follow.
Curator: Bora Pajo
Tomaino, G., & Jung, S. (2026). From people to prompts: Understanding susceptibility to persuasion in conversations with artificial intelligence. Journal of Personality and Social Psychology. Advance online publication. https://doi.org/10.1037/pspi0000531
Available here →November 1, 2026
20261101-tomaino-from-people-to-prompts-the-wide-angle
The wide angle

Geoff Tomaino is an Assistant Professor of Marketing at the University of Florida. He researches how consumers are adapting to an increasingly AI-embedded world. In addition to looking at how people are persuaded by AI, he also studies how consumers react to advertising in AI platforms, and how people judge the usefulness of these models.
People are more persuaded by AI because they think it cannot be persuaded, so they argue back less.
Most threads on Rorotoko begin with our invitation. Any thread can begin with a proposal: you write it with the same prompts every thread answers, and we review it.
Propose a thread →