MMPersuade：多模态说服数据集与评估框架

MMPersuade: A Dataset and Evaluation Framework for Multimodal Persuasion

October 26, 2025

作者: Haoyi Qiu, Yilun Zhou, Pranav Narayanan Venkit, Kung-Hsiang Huang, Jiaxin Zhang, Nanyun Peng, Chien-Sheng Wu

cs.AI

摘要

随着大型视觉语言模型（LVLM）在购物、健康、新闻等领域的广泛应用，它们正面临无处不在的 persuasive 内容。一个关键问题在于这些模型作为被说服者如何运作——即它们为何以及如何受到多模态 persuasive 输入的影响。理解模型对 persuasion 的易感性与不同 persuasive 策略的有效性至关重要，因为过度易受影响的模型可能采纳误导性信念、覆盖用户偏好，或在接触操纵性信息时生成不道德或不安全的输出。我们提出MMPersuade这一统一框架，用于系统研究LVLM中的多模态 persuasion 动态。该框架贡献包括：（i）一个综合多模态数据集，将图像和视频与商业、主观行为及对抗场景下的经典 persuasion 原则相匹配；（ii）基于第三方一致性评分和对话历史自估计 token 概率的评估框架，可量化 persuasion 有效性与模型易感性。通过对六种主流LVLM作为被说服者的研究，我们获得三项关键发现：（i）与纯文本相比，多模态输入显著提升 persuasion 有效性（及模型易感性），在错误信息场景中尤为突出；（ii）既存偏好声明虽降低易感性，但多模态信息仍保持 persuasion 优势；（iii）不同策略的效果因场景而异，互惠性在商业和主观场景中最有效，而可信度与逻辑性在对抗场景中占主导。通过联合分析 persuasion 有效性与模型易感性，MMPersuade为开发具有鲁棒性、偏好一致性且符合伦理规范的多模态 persuasion 应对模型奠定了理论基础。

English

As Large Vision-Language Models (LVLMs) are increasingly deployed in domains such as shopping, health, and news, they are exposed to pervasive persuasive content. A critical question is how these models function as persuadees-how and why they can be influenced by persuasive multimodal inputs. Understanding both their susceptibility to persuasion and the effectiveness of different persuasive strategies is crucial, as overly persuadable models may adopt misleading beliefs, override user preferences, or generate unethical or unsafe outputs when exposed to manipulative messages. We introduce MMPersuade, a unified framework for systematically studying multimodal persuasion dynamics in LVLMs. MMPersuade contributes (i) a comprehensive multimodal dataset that pairs images and videos with established persuasion principles across commercial, subjective and behavioral, and adversarial contexts, and (ii) an evaluation framework that quantifies both persuasion effectiveness and model susceptibility via third-party agreement scoring and self-estimated token probabilities on conversation histories. Our study of six leading LVLMs as persuadees yields three key insights: (i) multimodal inputs substantially increase persuasion effectiveness-and model susceptibility-compared to text alone, especially in misinformation scenarios; (ii) stated prior preferences decrease susceptibility, yet multimodal information maintains its persuasive advantage; and (iii) different strategies vary in effectiveness across contexts, with reciprocity being most potent in commercial and subjective contexts, and credibility and logic prevailing in adversarial contexts. By jointly analyzing persuasion effectiveness and susceptibility, MMPersuade provides a principled foundation for developing models that are robust, preference-consistent, and ethically aligned when engaging with persuasive multimodal content.