推理让大模型更诚实，因思维过程会引导其走向稳定的真实答案。
Think Before You Lie: How Reasoning Leads to Honesty
- 通过模拟道德权衡场景，发现推理能提升多类大模型的诚实度。
- 欺骗区域在表征空间中更易被扰动，而真实回答更稳定。
- 适合研究模型可解释性与伦理对齐的研究者阅读。
现有大语言模型评估多关注欺骗率，但其背后促成欺骗行为的条件仍不明确。本文构建了一个包含真实道德权衡的新数据集，其中诚实会带来可变成本。与人类在深思后反而更不诚实（Capraro, 2017; Capraro et al., 2019）不同，我们发现推理能持续提升多个规模和家族的大模型的诚实度。该效应并非仅由推理内容决定，因为推理轨迹常无法准确预测最终行为。相反，我们发现表征空间中的欺骗区域具有非稳态特性：相比诚实回答，欺骗输出更容易被输入改写、输出重采样及激活噪声所破坏。我们解释推理的作用机制为：生成推理令牌的过程实质上是在有偏的表征空间中遍历，最终将模型推向更稳定的诚实默认状态。
原文摘要 · Abstract (English)
While existing evaluations of large language models (LLMs) measure deception rates, the underlying conditions that give rise to deceptive behavior are poorly understood. We investigate this question using a novel dataset of realistic moral trade-offs where honesty incurs variable costs. Contrary to humans, who tend to become less honest given time to deliberate (Capraro, 2017; Capraro et al., 2019), we find that reasoning consistently increases honesty across scales and for several LLM families. This effect is not only a function of the reasoning content, as reasoning traces are often poor predictors of final behaviors. Rather, we show that the underlying geometry of the representational space itself contributes to the effect. Namely, we observe that deceptive regions within this space are metastable: deceptive answers are more easily destabilized by input paraphrasing, output resampling, and activation noise than honest ones. We interpret the effect of reasoning in this vein: generating deliberative tokens as part of moral reasoning entails the traversal of a biased representational space, ultimately nudging the model toward its more stable, honest defaults.
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。