Fig：自动化多智能体行为诱导器
原标题：Fig: Automated Multiagent Behavior Elicitor
AI 摘要
论文提出 Fig（Finding Interactions in Games），一个面向多智能体系统的自动化行为诱导系统。它接收自然语言描述的目标行为，将其分解为子行为，生成或选择社交博弈游戏，并对游戏引擎、对手、指令和上下文执行单项或组合干预，以诱导合作、欺骗、背叛等行为。在六种测试行为上，模型平均在 45.2% 的回合中尝试目标行为，相比匹配的对照游戏，尝试率平均提升 10.9%，表明多智能体环境可作为可控、可复现的行为评测平台。
正文节选
Fig: Automated Multiagent Behavior Elicitor Abstract Many social evaluations for multiagent systems study behaviors as they emerge from environments instead of systematically intervening in the environment to elicit such behaviors. We introduce Fig (Finding Interactions in Games), an agentic system that performs targeted elicitation interventions on game environments. It takes a natural language behavior description, decomposes it into sub-behaviors, generates new social games or selects existin