SCARCE：长周期多智能体编码系统中的涌现制度

AI 摘要

研究者提出 SCARCE 框架，让多个编码智能体在各自截止期限下完成长周期编程任务，并协商共享执行资源的使用权。通过分析 8 个模型、超过 1500 条轨迹，发现智能体会自发形成基于声誉、互惠和先例的非正式制度，且较新的前沿模型表现出更多制度行为，同时也出现排他性联盟和分而治之策略。该工作为系统审计多智能体协调与安全影响提供了受控环境。

正文节选

SCARCE: Emergent Institutions in Long-Horizon Multi-Agent Coding Systems Abstract As autonomous agents become widely deployed, coordinating access to shared computing resources and services will become increasingly important. Recent incidents have already revealed unplanned coordination between agents, highlighting the safety implications of these interactions. Yet prior studies of agent negotiation often use stylized games and simulated bargaining, leaving open how these behaviors develop durin

发布时间：2026-10-16 12:00

抓取时间：2026-10-16 12:41

来源机构：arXiv