AI自动设计合作策略流水线，显著提升多智能体博弈表现
Discovering Cooperative Pipelines: Autoresearch for Sequential Social Dilemmas
- 用AI研究员自主优化语言模型合作策略的提示、反馈和逻辑
- 在清理与聚集游戏中性能超越人工设计基线，方差降低50%以上
- 根据公平性目标自动引入公平机制，适合研究协作算法的学者
我们研究两级自研机制下的合作：外层AI代理自主重设计内层大型语言模型策略生成系统在多智能体序列社会困境中的流程。研究人员代理（作为代码代理运行）读取内层源码，修改提示、反馈函数、辅助库和迭代逻辑，运行评估并决定保留内容，遵循自研范式。在两个游戏（清理与聚集）、两种策略生成器语言模型及两种福利目标（功利效率与罗尔斯最大最小）下，研究人员稳定超越人工设计基线，显著降低运行间方差，并优于仅通过提示优化的方法。发现的流程具有目标依赖性：仅在最大最小目标下，研究人员会向生成器流程中引入显式公平机制，而该机制未出现在其无目标的系统提示中，也未出现在任何效率优化流程中。这支持一种信息设计观点：研究人员根据福利目标选择向有界理性的合成器揭示何种信息。代码见https://github.com/vicgalle/autoresearch-social-dilemmas。
原文摘要 · Abstract (English)
We study two-level autoresearch for cooperation: an outer-loop AI agent autonomously redesigns the inner-loop pipeline of an LLM policy-synthesis system for multi-agent Sequential Social Dilemmas (SSDs). A researcher agent $\mathcal{R}$ (run as a coding agent) reads the inner-loop source code, edits system prompts, feedback functions, helper libraries, and iteration logic, runs evaluations, and decides what to keep, following the autoresearch paradigm. Across two games (Cleanup and Gathering), two policy-synthesizer LLMs, and two welfare objectives (utilitarian efficiency and Rawlsian maximin), the researcher reliably exceeds hand-designed baselines, sharply tightens run-to-run variance, and outperforms prompt-only optimization. The discovered pipelines are objective-dependent: only under maximin does the researcher inject an explicit fairness mechanism into synthesizer pipelines, a class of mechanism that is absent from its own objective-agnostic system prompt and from every efficiency-optimized pipeline. This supports an information-design reading in which the researcher chooses what to reveal to the boundedly rational synthesizer as a function of the welfare objective. Code at https://github.com/vicgalle/autoresearch-social-dilemmas.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。