arXiv cs.AI· Yujiao Chen·· 9 天前AI 评分34
LLM 智能体为何偏袒本群体：代价、观察到的规范与声誉
Why LLM Agents Favor Their Group: Stakes, Observed Norms, and Reputation
AI 导读
研究用约 4400 个社会、330 万次模型调用测试 15 个 OpenAI 模型和 3 个 Claude 模型，发现单纯的群体标签效应只在给分不损己时存在，一旦有代价便消失。有代价时互动历史成为偏袒主因，15 个模型中 13 个呈统计显著正效应，11 个达 10 分中的约 3.5-8 分，并延伸至未谋面的标签陌生人。群体偏袒实为对观察到的群体行为的从众，会被个人声誉覆盖。
正文 · 原文
Abstract:Language-model agents favor their own group because they have watched their members favor each other. The group label alone does little once the decision has a cost; what drives favoritism is observed behavior, and an individual's own record can override it. We test this in small societies with arbitrary group labels, ten rounds of point sharing, and matched one-shot decisions across fifteen OpenAI models and three Claude models, about 4,400 societies and 3.3 million audited model calls. First, the large effect of a bare group label reported in earlier work appears only when giving others points costs the agent nothing; once the agent can keep points for itself, that effect collapses on every model that shows it. Second, under a stake, interaction history becomes the main source of favoritism: the history effect is statistically positive on 13 of 15 models, reaches about 3.5-8 points out of 10 on 11, grows with the number of rounds played, and extends to labeled strangers the agent has never met. Third, with scripted histories, favoritism falls to near zero under an egalitarian norm and reverses when the agent's own group is seen favoring the other side; stronger models side with an individual's record when it conflicts with the group. Group favoritism is thus conformity to observed group behavior, carried to strangers by the label and overridden by individual reputation. The same account predicts responses to betrayal, scandal, and a free offer to change group: public reprimand repairs betrayal better than apology or restitution, allocation punishment remains confined to the offending member, and a formed group cannot be bought but can, on weaker models, be invited away.
| Subjects: | Physics and Society (physics.soc-ph); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.11008 [physics.soc-ph] |
| (or arXiv:2610.11008v1 [physics.soc-ph] for this version) | |
| https://doi.org/10.48550/arXiv.2610.11008 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Yujiao Chen [view email]
[v1]
Wed, 7 Oct 2026 23:46:37 UTC (61 KB)
来源：arXiv cs.AI · arxiv.org