---
title: "交互税：多智能体团队中沟通如何抹除多样性"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2608.23541v1"
permalink: "https://feed.astralor.com/items/itm_9fc77a42cff048b5d9b5f527"
published: "2026-08-24T17:45:15.000Z"
score: 75
channel: "ai"
tags: ["LLM","交互税","多智能体系统","智能体协作"]
---

# 交互税：多智能体团队中沟通如何抹除多样性

## AI 点评

> 这篇论文值得所有搭建多智能体系统的人细读。它用一个轻量但却深刻的实验戳破了‘更多交互=更好结果’的直觉。对正在设计Agent框架的工程师来说，它的启示很直接：先独立采样，再决定要不要交互，且要谨慎控制智能体读到彼此完整答案。

## 速览

多智能体LLM交互究竟有益还是有害？论文指出，现有研究结论矛盾源于未区分交互类型。实验发现，不同模型家族会找到结构性不同的解，但一旦智能体读到彼此的完整输出，仅一轮就会收敛到同一方案，抹除了多模型协作所需的多样性，即“交互税”。在11个验证器打分优化任务、均衡预算下，全解交互是弱默认，独立生成可避免塌缩；批评仅在违规规则易被LLM发现和修复时有效。结论是多智能体表现更多取决于交换的信息，而非智能体数量。

## 关键点

- 提出“交互税”：完整信息交互导致多智能体方案在一轮内收敛，抹除多样性。
- 在11个验证器打分任务上，全解交互在均衡预算下弱于独立生成。
- 批评帮助有限，仅当违规规则容易定位和修复时有效。
- 多智能体性能的关键不是智能体数量，而是交换信息的类型与时机。

## 全文

Does multi-agent LLM interaction help or hurt? Some work reports gains from debate (Du et al., 2024), critique loops (Chen et al., 2025), and mixture-of-agents synthesis (Wang et al., 2025), while other work finds that interaction adds cost without improving quality under equal budgets (Tran & Kiela, 2026; Xu et al., 2026; Jarrett et al., 2025), or that independent sampling already captures multi-agent gains (Li et al., 2024). We argue this contradiction partly reflects a missing distinction, because not all multi-agent communication is equal. Different model families find structurally different solutions, but when agents read each other's complete outputs, their proposals converge within one round, erasing the diversity that motivates using multiple models. We call this the interaction tax. We test 11 verifier-scored optimization tasks under matched budgets and find that full-solution interaction is a weak default. Independent proposal generation avoids this collapse. Full-solution interaction mainly makes agents stay close to the first solution they see instead of trying different approaches, and critique helps only if the violated rule is easy for the LLM to find and fix. These results suggest that multi-agent performance depends less on the number of agents than on the information they exchange, and interaction helps only when agents share the right information at the right time.

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_9fc77a42cff048b5d9b5f527