强决策能力迫使智能体必须具备世界模型和信念记忆。
What Capable Agents Must Know: Selection Theorems for Robust Decision-Making under Uncertainty
- 通过赌局建模将预测任务转化为决策问题，推导出性能下限
- 高绩效要求具备世界模型、信念记忆及情绪类变量
- 适用于部分可观测和随机策略场景，无需假设最优或确定性
随着人工智能代理能力提升，其在不确定性下高效决策所需的核心结构是什么？经典理论表明最优控制可通过信念状态或世界模型实现，但未说明这些表示是必需的。本文证明了定量的“选择定理”：在低平均损失（低平均情形后悔）条件下，强制要求世界模型、类似信念的记忆结构，以及在任务混合下具有持续状态追踪功能的变量（类似于情感功能原语），并要求在块状结构任务中具备信息模块化。结果涵盖随机策略、部分可观测性和任务分布评估，不依赖最优性、确定性或显式模型假设。技术上，将预测建模简化为二元‘押注’决策，发现后悔边界限制了次优押注的概率质量，从而强制对高收益结果的区分能力。在完全可观测环境下，可近似恢复干预转移核；在部分可观测下，则表明预测状态与类信念记忆的必要性，解决了以往世界模型恢复研究中的开放问题。
原文摘要 · Abstract (English)
As artificial agents become increasingly capable, what internal structure is necessary for an agent to act competently under uncertainty? Classical results show that optimal control can be implemented using belief states or world models, but not that such representations are required. We prove quantitative "selection theorems" showing that strong task performance (low average-case regret) forces world models, belief-like memory and -- under task mixtures -- persistent regime-tracking variables resembling functional primitives of emotion, along with informational modularity under block-structured tasks. Our results cover stochastic policies, partial observability, and evaluation under task distributions, without assuming optimality, determinism, or access to an explicit model. Technically, we reduce predictive modeling to binary "betting" decisions and show that regret bounds limit probability mass on suboptimal bets, enforcing the predictive distinctions needed to separate high-margin outcomes. In fully observed settings, this yields approximate recovery of the interventional transition kernel; under partial observability, it implies necessity of predictive state and belief-like memory, addressing an open question in prior world-model recovery work.
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。