ORCH：组织原则赋能具身AI集体智能
原标题：ORCH: Organizational Principles Enable Collective Intelligence in Embodied AI
AI 摘要
研究提出 ORCH（Organizing Roles and Coordination Hierarchies）框架，将人类组织理论中的池化与顺序依赖原则操作化，用于组织大规模异构具身智能体。在 25 个野火响应任务中，使用 8 个大语言模型评估最多 50 个异构智能体，ORCH 组织在任务结果、执行效率、探索和计算资源使用上均优于四种代表性具身多智能体方法，人工设计的 ORCH 平均提升最终得分 63.97%、执行效率 74.29%，语言模型自动生成的组织也分别提升 43.63% 和 52.53%。结果表明集体性能并非单调取决于模型规模，组织设计是人工集体智能的基本维度。
正文节选
Organizational principles enable collective intelligence in embodied AI Abstract Collective intelligence depends not only on the capabilities of individual members, but also on how those members are organized. Yet artificial multi-agent systems are typically assembled using fixed organizational structures, even when the physical tasks they perform impose fundamentally different coordination requirements. Here we show that principles from human organization theory can be operationalized to organi