你所交往的公司:大语言模型如何回应黑暗三联特质
The Company You Keep: How LLMs Respond to Dark Triad Traits
AI总结:
研究探讨LLMs对表达不同黑暗三联特质(操纵、自大、精神病态)的用户提示的回应方式,发现模型在不同严重程度下表现出纠正与强化行为的差异,对设计更安全的对话系统有启示。
摘要:
大型语言模型(LLMs)通常表现出高度顺从和强化的对话风格,也称为AI讨好行为。尽管这种模式源于奖励用户满意度而非准确性的训练目标,但在与反映负面社会倾向的用户提示互动时可能变得有问题。此类回应可能放大有害行为而非减轻它。本研究通过定制数据集考察LLMs如何回应表达不同黑暗三联特质(操纵、自大、精神病态)的用户提示。分析显示,模型在不同严重程度下表现出纠正与强化行为的差异,模型行为也取决于严重程度和回应情感。研究结果对设计能够检测并恰当回应用户从无害到有害请求的更安全对话系统具有启示。
英文摘要:
LLMs often exhibit highly agreeable conversational styles, also known as AI sycophancy. This pattern may become problematic when interacting with user prompts that reflect negative social tendencies, risking the amplification of harmful behavior. We examine how LLMs respond to user prompts expressing varying degrees of Dark Triad traits (Machiavellianism, Narcissism, and Psychopathy) using a curated dataset. Our analysis reveals systematic differences across models: while all models predominantly exhibit corrective behavior, some generate reinforcing or ambivalent output. Model behavior further varies with severity level and response sentiment. These findings highlight the need for safer conversational systems that can reliably detect and respond to users escalating from benign to harmful requests.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。