通过信息网络引导多智能体 LLM 系统中的涌现行为
Steering Emergent Behavior in Multi-Agent LLM Systems via Information Networks
AI 导读
一篇 arXiv 论文（已被作者 Sepehr Ilami 撤回）研究了在多智能体 LLM 系统中如何设计观察网络来提升集体福利：10 个相同的 LLM 智能体玩重复群体决策博弈，管理者以固定预算添加定向观察链路，覆盖 1,080 个回合和六类博弈。结果显示信息价值取决于个体行为与集体目标的对齐程度，公共物品博弈中完全不共享信息反而最优，而模块化“杠铃”式连接最差。
正文 · 原文
This paper has been withdrawn by Sepehr Ilami
No PDF available, click to view other formats
Abstract:Large language model (LLM) agents increasingly operate in groups, each pursuing an individual objective while jointly determining system performance. If an operator controls which agent observes which other agent, but not their payoffs or messages, how should observations be structured to raise collective welfare? We make the information network (who observes whom) a design object, decoupled from the fixed interaction network over which agents act. Ten identical LLM agents play repeated group-decision games; by default each observes only an anonymous summary of the group's previous-round outcome, and a manager with a fixed budget adds directed observation links that reveal peers' past actions. Across 1,080 episodes and six games spanning coordination, anti-coordination, social dilemmas, and discovery, we ask how much information to share and, at a fixed budget, who should observe whom. Disclosure creates behavioral coupling: observation acts as a contagion channel whose welfare effect depends on the alignment between individually attractive behavior and the collective objective. First, the value of information is alignment-contingent: more peer information tends to help or be neutral when individual behavior is aligned with the collective goal, but under free-riding or congestion externalities it amplifies collectively harmful behavior; in a public-goods game the group does best with no information at all. We read this sign separation as a cross-game regularity within one controlled LLM population, not a model-invariant law. Second, at a fixed budget, who observes whom matters primarily when coupling is harmful: a modular "barbell" that lets sub-populations entrench harmful behavior is reliably the worst wiring, reversing the network-reciprocity intuition from human evolutionary game theory. These results argue against a default of full transparency for agent collectives.
| Comments: | It is preliminary and is not suitable for public release yet |
| Subjects: | Multiagent Systems (cs.MA); Artificial Intelligence (cs.AI); Computer Science and Game Theory (cs.GT); Social and Information Networks (cs.SI) |
| ACM classes: | I.2.11 |
| Cite as: | arXiv:2611.11714 [cs.MA] |
| (or arXiv:2611.11714v2 [cs.MA] for this version) | |
| https://doi.org/10.48550/arXiv.2611.11714 arXiv-issued DOI via DataCite |
Submission history
From: Sepehr Ilami [view email]
[v1]
Mon, 9 Nov 2026 00:18:26 UTC (277 KB)
[v2]
Tue, 10 Nov 2026 22:49:42 UTC (1 KB) (withdrawn)
来源：arXiv cs.AI · arxiv.org