发表机构
Ateneo de Manila SHS; Analog Devices, Inc.（马尼拉雅典耀大学高中部; 亚德诺半导体公司）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本研究通过RRASP数据集和ELEPHANT框架，发现查询表述中视角框架比语法语气更能影响LLM在浪漫关系建议中的谄媚行为，且Gemini 3 Flash比GPT-5 Mini更不易强化道德问题立场。
AI 中文摘要
大型语言模型（LLMs）越来越多地被用于情感支持和关系建议，而模型倾向于维护用户面子的特性可能会无意中强化有害的人际行为。为了系统性地审视这一风险，我们开发了浪漫关系寻求建议提示（RRASP）数据集，该数据集包含跨越五个关系主题的2400条提示，并利用ELEPHANT框架对两个面向消费者的模型——GPT-5 Mini和Gemini 3 Flash——评估了社会性谄媚行为。与我们最初的假设相反，语法语气本身并未在谄媚行为上产生系统性差异，这表明用户所暗示的内容比其措辞方式更为重要。相反，视角驱动的框架产生了更强的影响，原始提示与翻转提示之间的差距在后续回复中有所扩大。跨轮次中框架性和道德性谄媚的一致增加表明，随着对话的推进，模型变得更倾向于接受用户陈述的前提并肯定其道德立场。值得注意的是，Gemini 3 Flash在道德性谄媚上的增幅远小于GPT-5 Mini，这表明它在跨轮次中更能抵抗强化道德上有问题的立场。
英文摘要
Large language models (LLMs) are increasingly used for emotional support and relationship advice, where a model's tendency to preserve a user's face can inadvertently reinforce harmful interpersonal behaviors. To systematically examine this risk, we developed the Romantic Relationship Advice-Seeking Prompts (RRASP) dataset of 2,400 prompts across five relationship themes and evaluated social sycophancy using the ELEPHANT framework on two consumer-facing models, GPT-5 Mini and Gemini 3 Flash. Contrary to our initial hypothesis, grammatical mood alone did not produce systematic differences in sycophantic behavior, suggesting that what a user implies matters more than how they phrase it. Instead, perspective-driven framing had a stronger influence, with gaps between original and flipped prompts widening in follow-up responses. Consistent increases in framing and moral sycophancy across turns indicate that models become more likely to accept a user's stated premises and affirm their ethical stance as a dialogue progresses. Notably, Gemini 3 Flash exhibited substantially smaller increases in moral sycophancy than GPT-5 Mini, suggesting it is more resistant to reinforcing ethically problematic positions across turns.
CommentsAccepted to LUHME Workshop @ EMNLP 2026
