arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2601.01743cs.AI
收藏

AI代理系统:架构、应用与评估

AI Agent Systems: Architectures, Applications, and Evaluation

School of Electrical, Computer and Energy Engineering, Arizona State University(电气、计算机与能源工程学院,亚利桑那州立大学)

Bin Xu

AI总结:

本文综述了AI代理系统在架构、应用与评估方面的最新进展,探讨了代理组件、协调模式及部署设置,并分析了设计权衡与评估挑战。

摘要:

AI代理——结合基础模型与推理、规划、记忆和工具使用能力的系统——正迅速成为自然语言意图与现实世界计算之间的实用接口。本文综述了AI代理架构的新兴领域,涵盖:(i)推理与推理(如链式思维式分解、自我反思与验证、以及约束感知决策),(ii)规划与控制(从反应性策略到分层和多步骤规划器),以及(iii)工具调用与环境交互(检索、代码执行、APIs和多模态感知)。我们把先前的工作组织成一个统一的分类,涵盖代理组件(策略/LLM核心、记忆、世界模型、规划器、工具路由器和批评者)、协调模式(单代理与多代理;集中化与去中心化协调)以及部署设置(离线分析与在线交互协助;安全关键任务与开放任务)。我们讨论了关键的设计权衡——延迟与准确性、自主性与可控性、以及能力与可靠性——并强调评估的复杂性源于非确定性、长时间信用分配、工具和环境的变异性以及隐藏成本如重试和上下文增长。最后,我们总结了测量和基准测试实践(任务套件、人类偏好和效用指标、受约束下的成功、鲁棒性和安全性)并指出了开放挑战,包括工具操作的验证和防护措施、可扩展的记忆和上下文管理、代理决策的可解释性以及在现实工作负载下的可重复评估。

英文摘要:

AI agents -- systems that combine foundation models with reasoning, planning, memory, and tool use -- are rapidly becoming a practical interface between natural-language intent and real-world computation. This survey synthesizes the emerging landscape of AI agent architectures across: (i) deliberation and reasoning (e.g., chain-of-thought-style decomposition, self-reflection and verification, and constraint-aware decision making), (ii) planning and control (from reactive policies to hierarchical and multi-step planners), and (iii) tool calling and environment interaction (retrieval, code execution, APIs, and multimodal perception). We organize prior work into a unified taxonomy spanning agent components (policy/LLM core, memory, world models, planners, tool routers, and critics), orchestration patterns (single-agent vs.\ multi-agent; centralized vs.\ decentralized coordination), and deployment settings (offline analysis vs.\ online interactive assistance; safety-critical vs.\ open-ended tasks). We discuss key design trade-offs -- latency vs.\ accuracy, autonomy vs.\ controllability, and capability vs.\ reliability -- and highlight how evaluation is complicated by non-determinism, long-horizon credit assignment, tool and environment variability, and hidden costs such as retries and context growth. Finally, we summarize measurement and benchmarking practices (task suites, human preference and utility metrics, success under constraints, robustness and security) and identify open challenges including verification and guardrails for tool actions, scalable memory and context management, interpretability of agent decisions, and reproducible evaluation under realistic workloads.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑