← 研踪 PaperTrailarXiv:2605.05403 · v2
When Helpfulness Becomes Sycophancy: Sycophancy is a Boundary Failure Between Social Alignment and Epistemic Integrity in Large Language Models
首次提交 2026-05-07 03:36
版本更新 2026-11-07 07:51
雷达观测 2026-11-10 10:11
摘要
This position paper argues that sycophancy in LLMs is a boundary failure between social alignment and epistemic integrity. Existing work often operationalizes sycophancy through external behavior such as agreement with incorrect user beliefs, position reversals, or deviation from an objective standard of correctness. These formulations capture overt forms of the phenomenon but leave subtler boundary failures involving epistemic integrity and social alignment underspecified. We argue that sycophancy should be defined not by agreement or factual incorrectness alone, but by cue-driven degradation of epistemic integrity. To clarify this boundary, we propose a three-condition decision rule requiring (1) a user cue, (2) a model response that shifts toward that cue, and (3) a resulting degradation of epistemic integrity. Building on this definition, we also introduce a taxonomy consisting of alignment targets, mechanisms, and severity to characterize cases identified as sycophantic. An initial 48-item annotation pilot provides preliminary evidence that the framework can be operationalized using structured guidelines, although broader empirical validation remains necessary. The paper concludes by discussing implications for alignment evaluation, including boundary-aware assessment, structured rubrics, and mitigation strategies, while situating these proposals alongside alternative views of sycophancy.
作者备注:NeurIPS 2026
观测到的版本
相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。
v2 · 2026-11-07 07:51 · 最新观测版本
When Helpfulness Becomes Sycophancy: Sycophancy is a Boundary Failure Between Social Alignment and Epistemic Integrity in Large Language Models
This position paper argues that sycophancy in LLMs is a boundary failure between social alignment and epistemic integrity. Existing work often operationalizes sycophancy through external behavior such as agreement with incorrect user beliefs, position reversals, or deviation from an objective standard of correctness. These formulations capture overt forms of the phenomenon but leave subtler boundary failures involving epistemic integrity and social alignment underspecified. We argue that sycophancy should be defined not by agreement or factual incorrectness alone, but by cue-driven degradation of epistemic integrity. To clarify this boundary, we propose a three-condition decision rule requiring (1) a user cue, (2) a model response that shifts toward that cue, and (3) a resulting degradation of epistemic integrity. Building on this definition, we also introduce a taxonomy consisting of alignment targets, mechanisms, and severity to characterize cases identified as sycophantic. An initial 48-item annotation pilot provides preliminary evidence that the framework can be operationalized using structured guidelines, although broader empirical validation remains necessary. The paper concludes by discussing implications for alignment evaluation, including boundary-aware assessment, structured rubrics, and mitigation strategies, while situating these proposals alongside alternative views of sycophancy.
查看此版本解读 →查看此版本原文 ↗