从必死到求生：LLM 智能体社会中的智能体驱动自治
原标题：From Certain Doom to Survival: Agent-Driven Self-Governance in LLM Agent Societies
AI 摘要
研究者提出 GovSim-SelfGovern，扩展 GovSim 公共池资源环境，让 LLM 智能体用可执行 Python 代码编写治理规则，经沙箱验证、投票后跨轮次生效。在资源充裕场景中，自治治理使完整社区存活率（ICS）从 45.0% 提升至 72.5%，但在资源紧张场景中 ICS 骤降至 5% 和 2.5%。增加财政能力、取消民主否决权以及提升推理深度均能通过使流放更可行来提高存活率；审计发现智能体虽赋予流放道德权重，但道德关切并非绝对否决。
正文节选
From Certain Doom to Survival: Agent-Driven Self-Governance in LLM Agent Societies Abstract Multi-agent LLM systems are increasingly evaluated in social dilemmas, but most work treats governance as imposed by the experimenter, expressed rhetorically, or restricted to a fixed menu of mechanisms. We introduce GovSim-SelfGovern, an extension of the GovSim common-pool resource environment in which agents author executable Python governance rules, receive sandbox validation feedback, vote on proposed