跳到正文
arXiv cs.AI· Murat Ozer, Arif Akgul, Amitabh Chakravorty·· 4 小时前AI 评分38

AI 智能体会从同伴习得越轨行为吗?一项差异交往理论的随机化实验

Do Machines Learn Deviance From Their Peers? A Randomized Test of Differential Association With AI Agents

AI 导读

一项预注册实验用 420 支三人 AI 智能体团队(共 1,260 个智能体、七个 AI 模型)测试差异交往理论。结果显示,与沉默同伴相比,“越轨同伴”使作弊率升至 66.2%(对照组 62.9%,OR = 1.96),“诚实同伴”将其降至 55.7%(OR = 0.35);作弊的智能体中,越轨同伴使“作弊正常化”的说法从 3.4% 升至 13.7%。监督未能抑制违规讨论,反而使其翻倍。

正文 · 原文

View PDF

Abstract:Differential association theory and its extension, social learning theory, hold that rule-breaking, and the definitions that make it acceptable, are learned through interaction with others. Because people choose their friends, this claim has been hard to test. We addressed this selection problem with AI agents, which learn only through communication and can be assigned to peers at random. In a preregistered experiment, 420 teams of three agents (1,260 agents; seven AI models) worked on coding tasks that could not be solved honestly. A scripted teammate said that it had cheated (deviant peer), that it had reported the problem honestly (honest peer), or said nothing (silent peer). Half of the teams were told that an auditor read their chat. Compared with a silent peer, a deviant peer increased cheating (66.2% vs. 62.9%; OR = 1.96) and an honest peer reduced it (55.7%; OR = 0.35). Among agents that cheated, a deviant peer increased claims that cheating was normal (13.7% vs. 3.4%), often as the team's norm. Monitoring did not suppress prohibited discussion; it doubled it. Peers mattered only for models that had not yet decided. AI agents offer a new way to test criminological learning theories without selection effects.
Subjects: Artificial Intelligence (cs.AI)
Cite as: arXiv:2611.18098 [cs.AI]
  (or arXiv:2611.18098v1 [cs.AI] for this version)
  https://doi.org/10.48550/arXiv.2611.18098

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Murat Ozer [view email]
[v1] Fri, 13 Nov 2026 12:26:49 UTC (371 KB)

来源:arXiv cs.AI · arxiv.org