When AI Gets Persuaded, Humans Follow: Inducing the Conformity Effect in Persuasive Dialogue

TL;DR

The paper investigates whether a Persuadee Agent co-present in persuasive dialogue can induce conformity in a human participant and boost attitude change. It uses a between-subjects design with four conditions, leveraging GPT-4.1 to drive a triadic dialogue on healthy eating, and measures perceived persuasiveness, attitude change, and turn-by-turn dynamics. The findings show that a persuaded Persuadee Agent elevates both perceived persuasiveness and attitude change, especially when paired with an icebreaker, while an unpersuaded Persuadee Agent suppresses impact. These results reveal a practical and theoretical role for AI peers in persuasion, offering design guidelines for AI driven social influence in health and behavior change applications.

Abstract

Recent advancements in AI have highlighted its application in captology, the field of using computers as persuasive technologies. We hypothesized that the "conformity effect," where individuals align with others' actions, also occurs with AI agents. This study verifies this hypothesis by introducing a "Persuadee Agent" that is persuaded alongside a human participant in a three-party persuasive dialogue with a Persuader Agent. We conducted a text-based dialogue experiment with human participants. We compared four conditions manipulating the Persuadee Agent's behavior (persuasion acceptance vs. non-acceptance) and the presence of an icebreaker session. Results showed that when the Persuadee Agent accepted persuasion, both perceived persuasiveness and actual attitude change significantly improved. Attitude change was greatest when an icebreaker was also used, whereas an unpersuaded AI agent suppressed attitude change. Additionally, it was confirmed that the persuasion acceptance of participants increased at the moment the Persuadee Agent was persuaded. These results suggest that appropriately designing a Persuadee Agent can improve persuasion through the conformity effect.

Figures (19)

• Figure 1: Conceptual diagram of the persuasion process mediated by a Persuadee Agent. The upper section depicts the initial persuasion phase, where the Persuader Agent directs its message to both the Persuadee Agent and the Human Persuadee, eliciting skeptical reactions. The lower section represents later phases, where the Persuadee Agent's prior acceptance influences the Human Persuadee through a conformity effect, increasing their persuasion acceptance.
• Figure 2: The overall flow of the experiment. First, participants completed the "Evaluation Practice" and "Pre-Survey." Then, the dialogue session diverged according to the assigned experimental condition. After all dialogue sessions concluded, participants answered the "Post-Survey."
• Figure 3: Violin plot of overall perceived persuasiveness of the Persuader Agent (5-point scale).
• Figure 4: Violin plot of attitude change regarding participants' eating habits.
• Figure 6: Control