arXiv cs.AI· Burkay Tun\c{c}t\"urk, Koorosh Aslansefat, Yiannis Papadopoulos, Bao Nguyen·· 4 小时前AI 评分39
多智能体 LLM 更可靠吗？运行负载不确定性下的系统级可靠性分析
More Agents, More Reliable? System-Level Reliability of Multi-Agent LLMs under Operational-Profile Uncertainty
AI 导读
arXiv 论文提出一个基于 HIP-LLM 与贝叶斯网络的框架，评估多智能体 LLM 系统在运行负载不确定下的可靠性，将各智能体的失效概率区间传播为可靠性包络。在受监管化学品危险声明的提取流水线中，三个模型的多数投票能防护最弱模型，但未明显超过最佳模型，若一个投票模型在 5% 的请求上失效，其相对中间模型的优势将基本消失。结论是更多智能体带来更可辩护的可靠性论证，而非自动更可靠的系统。
正文 · 原文
Abstract:Teams increasingly build software from cooperating large language model (LLM) agents, often assuming more agents mean more reliability. Yet the usual evidence, average benchmark accuracy, ignores the input mix a deployed system meets and the rule merging the agents' answers. Building on HIP-LLM, a hierarchical imprecise-probability assessment of single LLMs, we propose a framework combining it with a Bayesian network of the system's failure logic for multi-agent systems. Agent failure-probability intervals propagate through this network, where the combining rule can itself fail, into a reliability envelope showing which evidence limits it. In a pipeline extracting regulated chemical-hazard statements, majority voting over three models protected against the weakest but did not clearly exceed the best, and a voter failing on 5\% of demands would largely erase its advantage over the middle model. More agents yield a more defensible reliability claim, not automatically a more reliable system.
| Comments: | Submitted for IEEE Software Magazine |
| Subjects: | Multiagent Systems (cs.MA); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2611.07140 [cs.MA] |
| (or arXiv:2611.07140v1 [cs.MA] for this version) | |
| https://doi.org/10.48550/arXiv.2611.07140 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Koorosh Aslansefat [view email]
[v1]
Wed, 4 Nov 2026 21:08:18 UTC (1,041 KB)
来源：arXiv cs.AI · arxiv.org