今日速览
| 序号 | 标题 | 来源 | 日期 | 主题 | 推荐等级 |
|---|---|---|---|---|---|
| 1 | TeamGym: A Gym for Team Decision-Making by AI Agents with Distributed Information | arXiv | 2026-10-27 | RAG | 高 |
| 2 | Gym-Anything-RT: Scaling Real-Time Environments for Vision-Language Agents | arXiv | 2026-10-27 | AI-Agent | 高 |
重点论文与技术动态
1. TeamGym: A Gym for Team Decision-Making by AI Agents with Distributed Information
- 来源：arXiv
- 日期：2026-10-27
- 作者/机构：Chanwoo Park, Jaywoong Jeong, Zachary Wojtowicz, Yubin Kim, Taekyun Lee, Kaiqing Zhang
- 主题标签：
RAG,arXiv - 推荐等级：高
- 分类：cs.AI
一句话结论
论文提出TeamGym环境，用于评估和强化学习训练大模型多智能体团队决策，量化揭示了通信与检索差距，并通过RL有效诱导出了协同行为。
核心内容
- 提出“关键信息”（决策充要最小事实集）概念，用于精准控制分布式信息并评估智能体的提取与整合能力。
- 评测揭示明显缺陷：信息分散导致成功率下降22.9%（通信差距），需自主检索关键信息再降19.9%（检索差距）。
- 仅依据最终决策结果进行强化学习，即可提升决策质量，并在发言选择、沟通策略等方面涌现出协作行为。
方法与数据
- 基于法律、金融、医疗和软件工程四大领域的12个真实语料库构建环境，对8个语言模型进行了评测与策略训练实验。
价值判断
- 值得关注：量化证明了当前大模型在分布式团队决策中存在严重的通信和检索瓶颈。
- 可复用点：“关键信息”基准构建方法，以及基于最终决策奖励的端到端团队RL训练范式。
- 局限/待核查：具体测试的大模型名称摘要未明确，复杂实际业务中的策略泛化性待验证。
摘要
Many real-world decisions require teams to integrate information distributed across participants. Yet AI agents are still primarily trained and benchmarked in isolation. We introduce TeamGym, an environment for evaluating and training language model-based agents on team decision-making problems constructed from real-world legal, finance, medical, and software engineering corpora. TeamGym introduces a method to identify a minimal set of facts, each element of which is necessary and all of which are collectively sufficient for a correct final decision, which we call pivotal information. This enables TeamGym to both control the distribution of pivotal information and track how well agents identify and integrate pivotal information when making team decisions. Examining 8 language models on 12 source corpora, we find that current AI agents often fail to effectively share information: Distributing pivotal information across participants lowers success rate by 22.9 percentage points on average relative to giving the same information to a single model (a communication gap). We also document that participants do not effectively retrieve decision-relevant information on their own: A team's success rate is lowered by a further 19.9 points when participants must identify and extract pivotal information from source material rather than receiving it directly (a retrieval gap). To help remedy these deficits, we use TeamGym to train policies through reinforcement learning on team decision loss. Across two policy-training experiments, we find that training improves final-decision quality and induces emergent coordination in participation, speaker selection, and communication, even though the reward evaluates only the final decision.2. Gym-Anything-RT: Scaling Real-Time Environments for Vision-Language Agents
- 来源：arXiv
- 日期：2026-10-27
- 作者/机构：Pranjal Aggarwal, Sean Welleck
- 主题标签：
AI-Agent,arXiv - 推荐等级：高
- 分类：cs.LG, cs.AI
一句话结论
论文提出 Gym-Anything-RT 框架，揭示了视觉语言智能体在动态实时交互场景中存在巨大能力短板。
核心内容
- 实时性能骤降：GPT-6 Astra (xHigh) 实时通过率仅 9.0%，而在环境暂停时可大幅提升至 75.0%。
- 出现绕过行为：具备写代码能力的前沿智能体会尝试编写代码，以规避实时感知与行动的限制。
- 开源评测基准：公开发布了环境生成框架及构建的 100 个交互小游戏以供后续研究。
方法与数据
- 基于 6 项实时交互能力分类体系，通过智能体管线（agentic pipeline）自动生成 100 个小游戏，支持在实时、减速或暂停状态下进行受控对比。
价值判断
- 值得关注：系统揭示了模型推理/前向传播期间因环境状态动态演变而造成的严重性能损失。
- 可复用点：开源的 Gym-Anything-RT 自动生成管线及 100 个支持时间调控的基准小游戏。
- 局限/待核查：摘要未明确 6 项能力分类的具体定义，也未详述所测试的其他智能体具体种类。