今日速览
| 序号 | 标题 | 来源 | 日期 | 主题 | 推荐等级 |
|---|---|---|---|---|---|
| 1 | SwarmWorld: Stigmergic technological evolution in societies of language-model agents | arXiv | 2026-08-26 | AI-Agent | 高 |
| 2 | ProgRouter: Online Progress-Guided Orchestration for Multi-Agent LLM Workflows under Quality-Cost Tradeoffs | arXiv | 2026-08-26 | RAG | 高 |
重点论文与技术动态
1. SwarmWorld: Stigmergic technological evolution in societies of language-model agents
- 来源：arXiv
- 日期：2026-08-26
- 作者/机构：Subhadeep Pal, Fiona Y. Wang, Markus J. Buehler
- 主题标签：
AI-Agent,arXiv - 推荐等级：高
- 分类：cs.AI, cond-mat.mtrl-sci, cs.CL
一句话结论
SwarmWorld中，无角色的LLM代理通过物理刺激（stigmergy）自组织形成技术社会，其技术组合比独立搜索更广泛、更稳健，优于强基准。
核心内容
- 代理在空间环境中自组织，无需预设角色或配方，探索资源、测试材料、构建持久人工制品，并编写可执行控制器，由确定性模拟器在代理移除后评估。
- 技术通过协作构建、可执行继承和持久代理-人工制品网络积累，重用主要始于物理观察而非通信。
- 代理分化为探索、构建、维护和协调行为，随世界成熟动态转变，形成更广技术组合。
方法与数据
- 代理在空间环境中探索，通过确定性模拟器评估控制器；摘要未明确具体数据集。
价值判断
- 值得关注：物理刺激机制支持技术社会自组织，无需直接通信，挑战传统多代理系统依赖对话的设计。
- 可复用点：物理刺激与代理-人工制品网络可复用于其他多代理系统以促进技术演化。
- 局限/待核查：功能效益依赖于结果和时间尺度，未明确长期可持续性。
摘要
Collective intelligence can emerge when individuals coordinate through a shared environment, allowing local actions to accumulate into durable social organization. Language-model agents offer a new substrate for this process, yet most multi-agent systems rely on direct conversation, predefined roles, or centralized workflows. It remains unclear whether decentralized agents can build functional technologies and outperform independent search. Here, initially homogeneous LLM agents in SwarmWorld self-organize without assigned roles or recipes into evolving technological societies. Agents explore a spatial environment, process resources, test materials, construct persistent artifacts, and write executable controllers evaluated by a deterministic simulator under unseen disturbances after the agents are removed. SwarmWorld splits cognition from consequence: agents propose architectures and controllers within fixed action and material schemas, while the simulated world determines function. Shared societies develop broader, more resilient technological portfolios than a strong best-of-N isolated-search baseline, although isolated search remains competitive for the strongest artifact. Agents differentiate into exploration, construction, maintenance, and coordination behaviors, transitioning as the world matures. Technologies accumulate through collaborative construction, executable inheritance, and persistent agent-artifact networks, with most reuse beginning through physical observation rather than communication. Explicit cultural mechanisms amplify collaboration and organization, but functional benefits depend on outcome and timescale. Physical stigmergy alone supports capable societies, while interaction drives persistent technological ecologies rather than universally superior individual inventions.2. ProgRouter: Online Progress-Guided Orchestration for Multi-Agent LLM Workflows under Quality-Cost Tradeoffs
- 来源：arXiv
- 日期：2026-08-26
- 作者/机构：Somgyuan Li, Ahmed M. Abdelmoniem, Shiqiang Wang
- 主题标签：
RAG,arXiv - 推荐等级：高
- 分类：cs.AI, cs.MA
一句话结论
ProgRouter通过在线进度引导路由框架，自适应选择LLM代理以平衡多智能体LLM工作流中的任务质量、时间预算与成本效率，显著降低操作成本并维持高性能。
核心内容
- 现有级联路由方法仅做单次查询级决策，无法适应多步工作流中任务进度、剩余难度及成本效率的动态变化。
- ProgRouter引入多视图任务进度评分器，融合粗粒度工作流结果与细粒度子任务完成信号、进度趋势及状态质量。
- 在HumanEval Plus（代码生成）、MBPP（代码）、MATH-500（数学推理）和ASQA（检索问答）数据集上，ProgRouter减少操作成本同时保持强任务解决性能。
方法与数据
- 方法核心为多视图进度评分器、双路径进度预测器及自适应元门控机制，用于实时估计进度增益。
- 实验数据集覆盖代码生成、数学推理和检索增强问答任务，摘要未明确具体数据量。
价值判断
- 值得关注：有效解决多智能体LLM工作流的成本-质量动态权衡问题，提升实际部署可行性。
- 可复用点：进度引导路由机制可迁移至其他需动态资源调度的AI工作流系统。
- 局限/待核查：摘要未明确提及具体局限性。