AI 代理的幻觉，靠“改历史”来治
大模型代理在长任务里会“带病工作”:它早先的假设已经错了、计划已经过时，却还留在上下文里，把后面的每一步都带偏。现有世界模型忙着预测工具会返回什么，可真实反馈就在手边，预测等于白费。这篇换了个思路：不预测未来，而是回头改历史——先给每个动作判型(关键、探索、噪音)，再把噪音推理和动作从同一段历史里重写掉，让代理基于修正后的状态继续走。在搜索、终端、软件工程三个领域训练，六个基准上比最强基线平均高 3.2 到 6.7 分。它不是你明天能用上的东西，但它指了个方向：与其让 AI 更会猜，不如让它更会忘。
📄 原文摘要(英文)
Recent advances in large language models (LLMs) have enabled agents to tackle long-horizon tasks across diverse environments. To further improve agent performance, existing language world models typically predict environment observations, yet reconstructing high-entropy, execution-dependent tool responses offers limited value when real feedback is available. Meanwhile, agents suffer from task-state contamination, where unsupported assumptions and outdated plans persist in history and distort subsequent decisions. We propose the Agent-Editing World Model (AEWM), which models how reasoning and actions shape future task progress rather than simulating tool responses. AEWM combines Action Judge to distinguish Critical, Exploratory, and Noisy decisions with State Revision to edit noisy reasoning--action continuations from the same observed history. EditAct integrates these capabilities with real execution, directly changing the state underlying subsequent decisions rather than merely providing critiques. We train AEWM across Search, Terminal, and Software Engineering through mid-training and supervised fine-tuning. AEWM achieves 70.5\% macro-F1 on our Action Judge benchmark, exceeding the strongest frontier baseline by 10.6 points. Across six benchmarks and three agent backbones, EditAct improves average scores by 3.2--6.7 points over the strongest baseline. Furthermore, rejection sampling fine-tuning on verified EditAct trajectories, termed AEWM-RFT, improves over Self-RFT by 2.2--2.6 points across three domains without online AEWM guidance.