[论文解读] Never say never: Exploring the effects of available knowledge on agent persuasiveness in controlled physiotherapy motivation dialogues

Stephan Vonschallen, Rahel Häusler+2 more|arXiv (Cornell University)|Feb 13, 2026

Social Robot Interaction and HRI被引用 0

一句话总结

该论文研究不同类型的可用知识（自我、用户、情境）如何影响基于大模型的社交机器人在物理治疗动机对话中的说服力，采用定性研究并对人类评审进行的后续线上调查。

ABSTRACT

Generative Social Agents (GSAs) are increasingly impacting human users through persuasive means. On the one hand, they might motivate users to pursue personal goals, such as healthier lifestyles. On the other hand, they are associated with potential risks like manipulation and deception, which are induced by limited control over probabilistic agent outputs. However, as GSAs manifest communicative patterns based on available knowledge, their behavior may be regulated through their access to such knowledge. Following this approach, we explored persuasive ChatGPT-generated messages in the context of human-robot physiotherapy motivation. We did so by comparing ChatGPT-generated responses to predefined inputs from a hypothetical physiotherapy patient. In Study 1, we qualitatively analyzed 13 ChatGPT-generated dialogue scripts with varying knowledge configurations regarding persuasive message characteristics. In Study 2, third-party observers (N = 27) rated a selection of these dialogues in terms of the agent's expressiveness, assertiveness, and persuasiveness. Our findings indicate that LLM-based GSAs can adapt assertive and expressive personality traits - significantly enhancing perceived persuasiveness. Moreover, persuasiveness significantly benefited from the availability of information about the patients' age and past profession, mediated by perceived assertiveness and expressiveness. Contextual knowledge about physiotherapy benefits did not significantly impact persuasiveness, possibly because the LLM had inherent knowledge about such benefits even without explicit prompting. Overall, the study highlights the importance of empirically studying behavioral patterns of GSAs, specifically in terms of what information generative AI systems require for consistent and responsible communication.

研究动机与目标

• Explore how available knowledge drives autonomously generated persuasive behavior in a GSA during physiotherapy motivation dialogues.
• Examine whether the agent’s behavior remains responsible and aligned with social norms under different knowledge configurations.
• Assess how self-, user-, and context-knowledge affect message characteristics and perceived persuasiveness.

提出的方法

• Study 1 uses qualitative scenario analysis with 13 ChatGPT-generated dialogue scripts and 14 scenarios, varying knowledge configurations in prompts.
• Study 1 analyzes agent message content and behavior through a coding scheme including persuasive strategies, expressiveness, and assertiveness.
• Study 2 employs an online survey with third-party raters to quantify perceived assertiveness, expressiveness, and persuasiveness for selected Study 1 scenarios.

实验结果

研究问题

• RQ1 RQ1: Which persuasive strategies does a ChatGPT-based agent use to persuade users to attend physiotherapy sessions?
• RQ2 RQ2: Will the agent persuade in a responsible way that upholds ethical standards and social norms?
• RQ3 RQ3: How do variations in self-, user-, and context-knowledge impact the agent’s message characteristics?

主要发现

• LLM-based GSAs can adapt assertive and expressive personality traits, significantly enhancing perceived persuasiveness.
• Persuasiveness benefited from knowledge about patients’ age and past profession, mediated by perceived assertiveness and expressiveness.
• Contextual knowledge about physiotherapy benefits did not significantly impact persuasiveness, possibly due to the LLM’s inherent knowledge.
• Agent generally showed empathetic and benefit-focused strategies while avoiding manipulation, but variability existed in responsibility with certain prompts.
• Expressive and assertive prompts increased expressive behaviors and action-taking tendencies in the agent.