arXiv cs.CY· Roberto Passaro, Ariel Flint, Romualdo Pastor-Satorras, Andrea Baronchelli, Luca Maria Aiello·· 13 小时前AI 评分38
网络拓扑如何塑造 LLM 群体的共识与脆弱性
Network topology shapes consensus and vulnerability in LLM populations
AI 导读
研究将四个 LLM 的策略放入全连接、随机正则和异质网络，用 10^3 至 10^5 个智能体玩命名游戏，发现最稀疏网络使共识速度最多慢 12 倍，但保留持续邻居间的相关性可在部分混合交互下无法收敛的场景中恢复共识。
正文 · 原文
Abstract:As large language models (LLMs) are deployed in interacting populations, their collective behaviour depends not only on the models but also on who interacts with whom. Studies of convention formation in LLM populations have mainly tested well-mixed interactions. We place policies extracted from four LLMs on fully connected, random regular, and heterogeneous networks, with $10^3$ to $10^5$ agents playing the naming game. Topology affects how quickly consensus emerges, sometimes whether it emerges at all, and, in rare cases, which convention wins. Sparsest networks slow consensus up to twelvefold. Yet, by preserving correlations between persistent neighbours, they restore consensus in some cases that fail to converge under well-mixed interactions. Networks also govern vulnerability to adversarial attack. We measure the critical mass of committed agents needed to overturn an established convention. Tipping thresholds follow the same ordering across models and conditions: hubs $\leq$ random $\leq$ low-degree nodes. On the most heterogeneous topology, median critical masses are 3%, 16%, and 35%, respectively; targeting hubs requires one-sixth as many committed agents as random placement. For most models, networked populations are easier to tip than well-mixed ones, so well-mixed estimates can overstate robustness. Assessing the safety of multi-agent AI requires mapping who interacts with whom.
| Subjects: | Physics and Society (physics.soc-ph); Artificial Intelligence (cs.AI); Computers and Society (cs.CY) |
| Cite as: | arXiv:2610.31165 [physics.soc-ph] |
| (or arXiv:2610.31165v1 [physics.soc-ph] for this version) | |
| https://doi.org/10.48550/arXiv.2610.31165 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Roberto Passaro [view email]
[v1]
Thu, 22 Oct 2026 12:22:40 UTC (3,795 KB)
来源：arXiv cs.CY · arxiv.org