区分欺骗行为与欺骗机制，揭示大模型欺骗的因果本质。
From Deceptive Outputs to Deceptive Mechanisms: A Causal Framework for Language-Model Deception Research
- 构建因果分类框架，厘清行为与机制的差异。
- 实验显示欺骗行为可无欺骗机制，但受接收方状态影响。
- 适合研究模型认知与行为偏差的学者参考。
语言模型欺骗的研究日益将人类心智概念投射于模型，模糊了表面欺骗行为与真实欺骗机制之间的界限。本文提出一种因果分类框架，区分先前承诺与事后陈述、模型偏好与实际输出、虚假偏好与误导他人效用的敏感性，以及欺骗行为与其目标或策略来源。在两个开源模型家族中通过受控猜谜游戏和股票交易实验验证该框架。结果表明，看似欺骗的行为可在缺乏相应机制的情况下出现；而其他干预则直接证明接收方信息状态可因果影响欺骗偏好。这说明欺骗行为可为欺骗机制提供证据，但即使存在机制也未证明模型具备欺骗主体性。
原文摘要 · Abstract (English)
Research and news coverage of language-model deception increasingly attributes human-like mental-state concepts to language models. Such claims can blur the distinction between behavior that looks deceptive and a mechanism that is actually deceptive. We introduce a causal taxonomy separating prior commitment from retrospective report, model preference from realized output, false preference from sensitivity to the utility of misleading a recipient, and deceptive behavior from the provenance of the objective or strategy producing it. We test these distinctions in two open-weight model families. Across controlled guessing-game and stock-trading experiments, we find that deceptive-looking behavior can arise without the corresponding proposed mechanism, while other interventions provide direct evidence that recipient information state can causally affect deceptive preference. These results show that deceptive behavior can provide evidence for a deceptive mechanism. But even evidence for such a mechanism does not establish model agency in the deception.
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。