arXiv cs.CL· Hazel Chen, Peter West, Hila Gonen·· 13 小时前AI 评分34
研究揭示语义泄漏驱动联想性性别偏见
Guilt by Association: Semantic Leakage Drives Associative Gender Bias
AI 导读
研究发现，即使移除提示词概念与特定实体的归属关系，语义泄漏依然存在，说明它是更广泛的联想倾向而非有缺陷的启发式。将这一发现扩展到性别偏见后，实验显示性别化概念仅出现在场景中（无需主体与之互动）就能以与主体主动接触时相近的程度将模型输出推向刻板印象。作者认为，语义泄漏驱动的这类偏见被现有评测所忽视。
正文 · 原文
Abstract:Semantic leakage occurs when a concept in the prompt steers model generation about an entity, even though the concept is irrelevant to it (e.g., a person who likes yellow is said to work as a school bus driver). It has mostly been studied in settings where the concept is directly attributed to the entity the model is asked about. This leaves open whether leakage is a flawed heuristic, where the model builds a profile of that entity, or a much broader associative tendency, e.g., working across seemingly unrelated entities. We argue for the latter: even after removing the attribution of the concept to the specific entity, leakage persists. Next, we extend this idea to gender bias. While work on gender bias in LMs typically looks at cases where the stereotyped concept is attributed to an entity, our findings suggest that a concept may steer outputs with its mere presence in the scene. We show this using gendered concepts under two conditions: when the subject engages with the concept, and when the concept just sits in the scene. We find that both shift model generation toward the stereotype to a similar extent. Together, our results suggest that semantic leakage drives a form of bias that is overlooked by current evaluations.
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.28944 [cs.CL] |
| (or arXiv:2610.28944v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.28944 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Hazel Chen [view email]
[v1]
Tue, 20 Oct 2026 21:46:27 UTC (441 KB)
来源：arXiv cs.CL · arxiv.org