让大模型在生成时自我检查，实时纠正偏航行为
Cognition-of-Thought Elicits Social-Aligned Reasoning in Large Language Models
- 引入认知自检框架，在推理时动态监控生成过程
- 检测到潜在风险后回滚并注入社会共识引导重生成
- 无需重训练即可灵活调整安全策略，适合高风险应用
大语言模型虽具备复杂推理能力，但仍可能产生有害行为。现有对齐方法通常将安全规则嵌入模型权重，导致控制隐式、静态且难以修改。本文提出认知-思维（Cognition-of-Thought, CooT）框架，一种解码时的显式认知自检机制。该框架将标准文本生成器与一个持续监控生成序列的认知感知器相结合。感知器采用基于优先级的结构化原则层级（如安全高于服从），实时识别潜在偏差。一旦发现违规，CooT 将回滚至出错点，并在注入通用社会先验与上下文特定警告的指导下重新生成。该方法将对齐从固定属性转变为推理期间可显式、动态、可审计的过程，支持无需重训练的策略灵活更新。多基准测试和多种模型族的实验表明，CooT 能持续提升安全性和社会推理表现。
原文摘要 · Abstract (English)
Large language models (LLMs) excel at complex reasoning but can still exhibit harmful behaviors. Current alignment strategies typically embed safety into model weights, making these controls implicit, static, and difficult to modify. This paper introduces Cognition-of-Thought (CooT), a novel decoding-time framework that equips LLMs with an explicit cognitive self-monitoring loop. CooT couples a standard text Generator with a cognitive Perceiver that continuously monitors the unfolding sequence. The Perceiver uses a structured, precedence-based hierarchy of principles (e.g., safety over obedience) to detect potential misalignments as they arise. When violations are flagged, CooT intervenes by rolling back the generation to the point of error and regenerating under injected guidance that combines universal social priors with context-specific warnings. CooT thus transforms alignment from a fixed property into an explicit, dynamic, and auditable process active during inference, allowing for flexible policy updates without retraining the model. Extensive experiments across multiple benchmarks and model families confirm that CooT consistently improves safety and social reasoning performance.
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。