操纵与否：LLM 交互中的意见动力学
原标题：To Manipulate or Not: Opinion Dynamics in LLM Interactions
AI 摘要
Temple University 的研究者提出一个受控的 LLM–LLM 辩论框架，让两个语言模型角色围绕气候变化归因话题进行 20 轮对话，其中一方被赋予立场并可能被指示去说服对方。研究使用 GPT-6 Luna 和 Gemini 2.5 Flash Lite 组成四种模型配对，同时追踪双方自报立场和对话内行为。结果显示，同模型配对平均缩小意见差距，跨模型配对则扩大差距，且明确的操纵指令并不能稳定提升影响力。
正文节选
To Manipulate or Not: Opinion Dynamics in LLM Interactions 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 inst