arXiv cs.CL· Rhitabrat Pokharel, Ameeta Agrawal·· 13 小时前AI 评分42
多语言 LLM 的鲁棒性与谄媚性如何区分：SC-matrix 框架
Disentangling Robustness and Sycophancy in Multilingual LLMs
AI 导读
研究者提出 SC-matrix，将多语言 LLM 在无压力下能否答对（能力）与受压后能否坚持答案（谄媚性）交叉评估，以区分模型附和用户错误前提是出于谄媚还是该语言能力不足。结果显示，相同的总体谄媚率可能掩盖不同的失败构成，某语言上的高能力并不总能防止谄媚，且推理导向的缓解策略优于指令式策略。多语言谄媚并非单一现象，而是需要分别诊断的多种失败复合体。
正文 · 原文
Abstract:When a multilingual LLM agrees with a user's false premise, is it being sycophantic, or does it simply not know better in that language? Current research does not distinguish these cases: they record only whether the model agreed, not why. We introduce the SC-matrix, that crosses a model's competence (whether it answers correctly under no pressure) with its sycophancy (whether it maintains that answer under pressure). Our results show that (1) identical aggregate sycophancy rates can mask different failure compositions, (2) high competence in a language does not always protect against sycophancy, and (3) reasoning-oriented mitigations outperform directive-based strategies. We find that multilingual sycophancy is not a single phenomenon but a composite of different failures that require distinct diagnosis.
| Subjects: | Artificial Intelligence (cs.AI); Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.30696 [cs.AI] |
| (or arXiv:2610.30696v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.30696 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Rhitabrat Pokharel [view email]
[v1]
Thu, 22 Oct 2026 04:10:03 UTC (717 KB)
来源：arXiv cs.CL · arxiv.org