East China Normal University(华东师范大学);Shanghai Innovation Institute(上海创新研究院);Harbin Institute of Technology(哈尔滨工业大学);Xiamen University(厦门大学);Shanghai Artificial Intelligence Laboratory(上海人工智能实验室);Independent Researcher(独立研究者)
Deep research agents (DRAs) integrate LLM reasoning with external tools. Memory systems enable DRAs to leverage historical experiences, which are essential for efficient reasoning and autonomous evolution. Existing methods rely on retrieving similar trajectories from memory to aid reasoning, while suffering from key limitations of ineffective memory evolution and increasing storage and retrieval costs. To address these problems, we propose a novel Memory Intelligence Agent (MIA) framework, consisting of a Manager-Planner-Executor architecture. Memory Manager is a non-parametric memory system that can store compressed historical search trajectories. Planner is a parametric memory agent that can produce search plans for questions. Executor is another agent that can search and analyze information guided by the search plan. To build the MIA framework, we first adopt an alternating reinforcement learning paradigm to enhance cooperation between the Planner and the Executor. Furthermore, we enable the Planner to continuously evolve during test-time learning, with updates performed on-the-fly alongside inference without interrupting the reasoning process. Additionally, we establish a bidirectional conversion loop between parametric and non-parametric memories to achieve efficient memory evolution. Finally, we incorporate a reflection and an unsupervised judgment mechanisms to boost reasoning and self-evolution in the open world. Extensive experiments across eleven benchmarks demonstrate the superiority of MIA.
Emergent collective intelligence from massive-agent cooperation and competition
Tsinghua University, Shenzhen International Graduate School(清华大学深圳国际研究生院);University of California, San Diego(加州大学圣地亚哥分校);Georgia Institute of Technology(佐治亚理工学院)
Can Agentic Trading Systems Pay for Their Own Intelligence?
HKUST(GZ)(香港科技大学(广州));Paradoox AI(悖论人工智能公司);E Fund Management Co., Ltd(易方达基金管理有限公司);MBZUAI(Mohamed Bin Zayed University of Artificial Intelligence(穆罕默德·本·扎耶德人工智能大学));The University of Tokyo(东京大学)
Open Ended Intelligence: The individuation of Intelligent Agents
The Global Brain Institute, VUB(全球大脑研究所,布鲁塞尔自由大学)
通用人工智能(Artificial General Intelligence)是一个研究领域,旨在提取独立于特定问题域或预定义上下文的智能原理,并利用这些原理来合成能够执行人类能够执行的任何智力任务的系统,并最终超越人类。虽然专注于解决特定问题的“狭义”人工智能,如语音识别,文本理解,视觉模式识别,机器人运动等,最近已经显示出相当多的令人印象深刻的突破,理解一般智能仍然难以捉摸。在本文中,我们提供了一个新的理论方法来理解一般智力。我们首先简要介绍目前的概念方法。我们的批判暴露了一些严重的局限性,这些局限性可以追溯到智力概念的本体论根源。然后,我们提出了一个范式的转变,从智能感知为一个能力的个体代理定义的先验给定的问题域或目标,智能感知为一个形成过程的自组织智能代理个性化。我们称之为开放式智能。开放式智能是作为认知发展过程的抽象而发展起来的,因此它的应用可以扩展到一般的智能体和系统。我们介绍并讨论了这一思想的三个方面:个性化的哲学概念,意义的制作和一般认知主体的个性化。我们进一步展示了如何在一个分布式的,自组织的网络相互作用的元素,以及如何这样的过程是可扩展的开放式智能的框架。该框架强调了协调与智慧之间的重要关系以及对价值观的新理解。最后,我们提出了一些问题,供今后的研究。
AINTMA: Agentic AI Architecture for Autonomous Test Management with Generative Intelligence, Secure Cloud Communication and Adaptive Quality Analytics
International Business Machines (IBM)(国际商业机器公司(IBM));Global Atlantic Financial(全球大西洋金融公司);Docusign(DocuSign公司);Salesforce Inc(Salesforce公司);The Home Depot(家得宝公司)
Paradigms of Intelligence Team, Google(谷歌智能范式团队);University of Chicago(芝加哥大学);Santa Fe Institute(圣菲研究所);Antikythera, Berggruen Institute(安提基特拉与伯格格林研究所);University of California, San Diego(加州大学圣地亚哥分校)
Bio AI Agent: A Multi-Agent Artificial Intelligence System for Autonomous CAR-T Cell Therapy Development with Integrated Target Discovery, Toxicity Prediction, and Rational Molecular Design
Bio LIMS INC(Bio LIMS公司)
针对CAR-T疗法开发周期长、临床损耗率高的问题,提出多智能体系统Bio AI Agent,通过六个专业智能体协作自主完成靶点选择、毒性预测等CAR-T开发全流程,回顾性验证效果显著,可加速免疫疗法转化。
Polycraft World AI Lab (PAL): An Extensible Platform for Evaluating Artificial Intelligence Agents
The University of Texas at Dallas(德克萨斯大学达拉斯分校);Georgia Institute of Technology and Emory University(佐治亚理工学院和埃默里大学)
随着人工智能研究的进步,用于评估人工智能智能体的平台需要适应和发展,以继续挑战它们。我们介绍了Polycraft World AI Lab(PAL),这是一个基于Minecraft mod Polycraft World的API任务模拟器。我们的平台旨在允许具有不同架构的人工智能智能体轻松地与Minecraft世界进行交互,训练并在多个任务中进行评估。PAL允许以灵活的方式创建任务,并能够在评估期间操纵任务的任何方面。人工智能智能体和外部参与者(非玩家角色,NPC)在开放世界环境中采取的所有行动都被记录下来,以简化评估。在这里,我们提出了两个自定义任务的PAL平台上,一个专注于多步规划和一个专注于导航,并评估代理解决这些问题。总之,我们报告了一个多功能和可扩展的人工智能评估平台,人工智能研究人员可以利用它的低门槛。
Adaptive AI Agent Placement and Migration in Edge Intelligence Systems
Institute of Artificial Intelligence and Future Networks, Beijing Normal University, China(人工智能与未来网络研究院,北京师范大学,中国);Faculty of Arts and Sciences, Beijing Normal University, China(文理学院,北京师范大学,中国);Department of Computer Science and Engineering, Shanghai Jiao Tong University, China(计算机科学与工程系,上海交通大学,中国);College of Information Engineering, Zhejiang University of Technology, China(信息工程学院,浙江工业大学,中国)
MAGMA: A Multi-Graph based Agentic Memory Architecture for AI Agents
Department of Computer Science, The University of Texas at Dallas(德克萨斯大学达拉斯分校计算机科学系);Department of Electrical and Computer Engineering, University of Florida(佛罗里达大学电气与计算机工程系)
Deep Reinforcement Learning Agents are not even close to Human Intelligence
Department of Computer Science, Technical University Darmstadt(德累斯顿技术大学计算机科学系);National Research Center for Applied Cybersecurity (ATHENE)(应用网络安全国家研究中心(ATHENE));Hessian Center for Artificial Intelligence (hessian.AI)(黑森人工智能中心(hessian.AI));Centre for Cognitive Science, Darmstadt(达姆施塔特认知科学中心);Institute of Psychology, Technical University Darmstadt(德累斯顿技术大学心理学系);German Research Center for Artificial Intelligence (DFKI)(德国人工智能研究中心(DFKI));Sorbonne Université, Paris, France(巴黎索邦大学)
Towards 6G Native-AI Edge Networks: A Semantic-Aware and Agentic Intelligence Paradigm
College of Computer Science, University of Exeter, Exeter, U.K.(埃克塞特大学计算机科学学院);Sony China Research Laboratory, Beijing, China(索尼中国研究实验室);School of Information Science and Engineering, Southeast University, Nanjing, China(东南大学信息科学与工程学院);Purple Mountain Laboratories, Nanjing, China(紫金山实验室);Information System Technology and Design Pillar, Singapore University of Technology and Design, Singapore(新加坡科技设计大学信息系统技术与设计支柱)
Arena: A General Evaluation Platform and Building Toolkit for Multi-Agent Intelligence
University of Oxford(牛津大学);Lighthouse(灯塔公司);Beihang University(北京航空航天大学);Hebei University of Technology(河北工业大学);Imperial College London(帝国理工学院);University of California, Los Angeles(加利福尼亚大学洛杉矶分校)
AgentArk: Distilling Multi-Agent Intelligence into a Single LLM Agent
Carnegie Mellon University(卡内基梅隆大学);William & Mary(威廉与玛丽学院);Georgia Institute of Technology(佐治亚理工学院);Amazon(亚马逊);University of British Columbia(不列颠哥伦比亚大学)