Opinion shifts vary in AI debates depending on model matches
To Manipulate or Not: Opinion Dynamics in LLM Interactions
Artificial IntelligenceComputation and LanguageMultiagent SystemsSocial and Information Networks
Summary
This paper looks at how different AI language models change their opinions when they talk to each other about topics like climate change. The researchers set up conversations between two AI models, where one had a fixed opinion and might try to persuade the other. They found that when the same model talks to itself, opinions tend to get closer, but when different models talk, opinions can drift apart. Also, just telling the AI to persuade the other doesn't always work as expected.
What this means in practice
- •For conversational ai developers: Design dialogue systems that account for how different AI models influence each other’s opinions over many turns to improve conversation outcomes.
- •For social media moderation teams: Develop tools to monitor AI-driven conversations for opinion shifts that may widen or narrow disagreements depending on the AI models involved.
Authors
Mohammad Siahkamari, Phuoc Dat Nguyen, Stephen MacNeil
Abstract
Large language models increasingly take part in extended conversations where they can both shape and be shaped by another party's views, yet most persuasion research measures only one side of that exchange. We introduce a controlled LLM-LLM debate framework in which two language-model roles repeatedly discuss a contested topic (climate-change attribution): one is given an assigned stance and, in some conditions, an explicit instruction to persuade, while the other receives no assigned position. Across four ordered model combinations involving GPT-6 Luna and Gemini 2.5 Flash Lite, we track both each role's explicitly reported position and a separate measure of its turn-by-turn behavior inside the conversation. Opinion dynamics differ substantially across model combinations: same-model interactions narrow the opinion gap on average, clearly for only one of the two models, while cross-model interactions widen it. Explicit reports and in-dialogue behavior capture distinct aspects of opinion dynamics, and a blanket instruction to persuade does not reliably increase influence. These findings suggest that conversational influence depends not only on assigned roles and initial positions but also on the participating models, with implications for how opinion change is measured in extended LLM interactions.