思维认知激发大语言模型中的社会对齐推理
Cognition-of-Thought Elicits Social-Aligned Reasoning in Large Language Models
AI总结：
本文提出思维认知（CooT）解码框架，通过认知感知器监控生成过程并干预违规，将安全对齐转为动态可审计过程，显著提升模型安全性与社会推理能力。
摘要：
大语言模型（LLMs）在复杂推理方面表现出色，但仍可能展现出有害行为。当前的对齐策略通常将安全性嵌入模型权重中，使得这些控制隐式、静态且难以修改。本文引入了思维认知（CooT），一种新颖的解码时框架，为LLMs配备了显式的认知自我监控循环。CooT将标准文本生成器与认知感知器耦合，后者持续监控正在生成的序列。感知器使用基于优先级的结构化原则层级（例如，安全高于服从）来在潜在错位出现时进行检测。当标记出违规时，CooT通过将生成回滚到错误点，并在注入的指导下重新生成来进行干预，该指导结合了通用社会先验与上下文特定警告。因此，CooT将对齐从固定属性转变为推理过程中显式、动态且可审计的过程，允许在不重新训练模型的情况下进行灵活的策略更新。跨多个基准和模型家族的广泛实验证实，CooT持续提高了安全性和社会推理性能。
英文摘要：
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.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
