提出三种智能体控制架构，平衡自主性与可监管性
The Cartesian Cut in Agentic AI
- 将大模型预测与运行时系统分离，通过符号接口实现控制外化
- 对比三种控制模式：受限服务、笛卡尔智能体、集成智能体
- 适用于需要可监管性的高风险自动化场景
大语言模型通过预测人类文本获得能力，常反映人类执行任务的方式。将大模型与工程化运行时结合，可将预测转化为控制：输出触发干预以实现目标导向行为。我们指出，这类系统中控制的位置是核心设计杠杆。大脑将预测嵌入分层反馈控制器，由行动后果校准；而大模型智能体采用笛卡尔式代理：学习核心与工程化运行时通过符号接口连接，外部化控制状态与策略。这种分离带来自举、模块化与治理优势，但也可能引发敏感性和瓶颈问题。本文概述了受限服务、笛卡尔智能体与集成智能体三类控制范式，它们在自主性、鲁棒性与监督之间权衡。
原文摘要 · Abstract (English)
LLMs gain competence by predicting words in human text, which often reflects how people perform tasks. Consequently, coupling an LLM to an engineered runtime turns prediction into control: outputs trigger interventions that enact goal-oriented behavior. We argue that a central design lever is where control resides in these systems. Brains embed prediction within layered feedback controllers calibrated by the consequences of action. By contrast, LLM agents implement Cartesian agency: a learned core coupled to an engineered runtime via a symbolic interface that externalizes control state and policies. The split enables bootstrapping, modularity, and governance, but can induce sensitivity and bottlenecks. We outline bounded services, Cartesian agents, and integrated agents as contrasting approaches to control that trade off autonomy, robustness, and oversight.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。