PulseAugur
实时 08:23:01
(CA) Moral Hazard in Multi-Agent Language Models

新博弈测试多智能体语言模型中的合作失败

研究人员开发了一个新框架——对话道德风险博弈(Dialogue Moral Hazard Game),用于研究多智能体语言模型中的合作失败。该博弈实现了Holmström的团队道德风险模型,其中智能体可以承担成本来揭示对他人有利的信息。对七个开源模型的评估表明,许多智能体优先考虑即时的局部奖励,而不是有成本的信息共享,即使经过优化,团队成功的改进也并不总是与预期的合作机制相关联。 AI

影响 强调了在多智能体AI中进行机制层面评估的必要性,而不仅仅是团队成功。

排序理由 学术论文,介绍了一种新的多智能体语言模型框架和评估方法。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.MA (Multiagent) 阅读 →

AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →

新博弈测试多智能体语言模型中的合作失败

报道来源 [2]

  1. arXiv cs.AI TIER_1 (CA) · Dane Malenfant ·

    Moral Hazard in Multi-Agent Language Models

    arXiv:2607.23982v1 Announce Type: cross Abstract: Cooperation can fail when socially valuable effort is costly, weakly observable, and mainly benefits others. Drawing on Holmstr\"om's team moral-hazard model, we introduce the Dialogue Moral Hazard Game, a controlled textual game …

  2. arXiv cs.MA (Multiagent) TIER_1 (CA) · Dane Malenfant ·

    多智能体语言模型中的道德风险

    Cooperation can fail when socially valuable effort is costly, weakly observable, and mainly benefits others. Drawing on Holmström's team moral-hazard model, we introduce the Dialogue Moral Hazard Game, a controlled textual game that operationalizes this hidden-action structure fo…