arXiv:2609.27038v1 Announce Type: new Abstract: Chain-of-thought (CoT) monitoring assumes that the reasoning a model writes reflects the computation that directly produces its answer. Previous faithfulness metrics have been predominantly behavioral, as they simply edit the reasoning text and observe the resulting answer. However, our methodology aims to measure faithfulness causally at the activation level, specifically on self-generated reasoning....
展望未来,AI 监测与解释的方向将不再局限于文本层面。研究人员可能需要转向分析模型的内部状态激活模式、注意力权重分布以及中间层的数值变化。只有深入到“黑盒”内部,捕捉真正决定输出的计算节点,我们才能构建出真正可靠的 AI 系统。
此外,这也提醒我们在设计提示词和交互界面时要保持谨慎。用户不应仅仅依赖模型口述的“思考过程”来判断其可靠性,而应将其视为一种辅助的、可能不完全准确的解释。未来的 CoT 研究或许会朝着“可控性”和“可追踪性”双重目标发展,即在保证答案正确的同时,提供更真实的计算轨迹。这篇论文的出现,无疑为 AI 可解释性领域吹响了集结号,也提醒我们:在追求透明的道路上,或许需要打破既有的认知惯性。