语言模型缺持久状态追踪，世界模型或更适长程推理
Why We Need World Models for AGI: Where LLMs Fail and How World Models May Outperform
- 将文本规则转为隐式动态系统，构建可解析的推理环境
- 显式状态空间代理胜率79%，语言模型仅11%，差距显著
- 适合研究长期规划与因果推理的模型改进方向
大型语言模型在语言生成和知识密集型任务中表现优异，但在需要因果推理、持续状态追踪和长程规划的场景中仍受限。我们认为，这源于序列预测目标与潜环境动态推理之间的本质不匹配。为此提出潜动态推断（LDI）概念框架，将语言和多模态观测视为潜在转移动态的部分证据。为实证检验该观点，构建了全由自然语言规则定义的顺序推理环境Flux。作为概念验证，先将规则编译为显式状态转移模拟器，证明部分潜转移动态可从文本规则中提取。由此可在可控环境下对比仅依赖文本观测的LLM与直接在提取的潜状态空间中训练的强化学习代理。结果显示，拥有显式状态访问权限的代理在长程游戏中行为更稳定，总胜率约79%，而LLM仅为11%。定性分析揭示其失败模式集中于状态追踪不稳定，如无效动作、状态错误和短程推理。完整Flux环境实现见https://github.com/FeisalAlaswad/FLUX-RL-Agent。在该设定下，仅靠强序列预测难以支撑鲁棒的长程动态推理，缺乏持续状态追踪与转移建模机制。
原文摘要 · Abstract (English)
Large language models achieve strong performance in language generation and knowledge-intensive tasks, yet remain limited in settings requiring causal reasoning, persistent state tracking, and long-horizon planning. We argue that these limitations may arise from an objective-level mismatch between sequence prediction and reasoning over latent environment dynamics. To formalize this distinction, we introduce Latent Dynamics Inference (LDI), a conceptual perspective that interprets language and multimodal observations as partial evidence of underlying transition dynamics. To empirically investigate this perspective, we introduce Flux, a sequential reasoning environment specified entirely through natural-language rules. As a proof-of-concept case study, the rules are first compiled into an explicit state-transition simulator, illustrating that structured latent transition dynamics can, in some cases, be operationally extracted from textual rule descriptions. This enables a controlled comparison between the LLMs operating purely over textual observations and reinforcement-learning agents trained directly within the extracted latent state space. Within this case study, agents operating with explicit access to the latent state space exhibit substantially more stable behavior in long-horizon gameplay, achieving an aggregate win rate of approximately 79% versus 11% for LLMs. Qualitative analysis further reveals failure modes consistent with unstable persistent state tracking, including invalid actions, state-tracking errors, and short-horizon reasoning failures. The complete implementation of the Flux environment available at https://github.com/FeisalAlaswad/FLUX-RL-Agent Within the evaluated setting, these results suggest that strong sequence prediction alone may struggle to support robust long-horizon dynamic reasoning without mechanisms for persistent state tracking and transition modeling
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。