语言模型中的自我纠正并非一种固有能力
Self-correction is Not An Innate Capability in Language Models
AI总结：
本文研究语言模型是否具备固有道德自我纠正能力，通过行为和机制分析发现其缺乏道德敏感性和有效整合外部反馈的能力。
摘要：
尽管对大型语言模型（LLMs）的自我纠正能力已有越来越多的兴趣，但关于其有效性的结论却存在差异。先前的研究大多集中在内在自我纠正和外在自我纠正上，特别是内部知识与外部反馈之间的相互作用仍处于探索阶段。在本文中，我们旨在全面研究道德自我纠正的底层机制，通过回答一个基本问题：道德自我纠正是否是LLMs的固有能力？具体来说，我们进行了：（1）基于自我区分任务的LLMs道德敏感性行为分析；以及（2）对隐藏状态的机制分析，以考察自我纠正的关键组成部分，如链式思维（CoT）和外部反馈，如何相互作用以促进道德自我纠正。基于行为和机制分析的实证证据，我们证明道德自我纠正并非LLMs的固有能力，因为它们既不具有道德敏感性，也未能在自我纠正过程中有效整合外部反馈。
英文摘要：
Although there has been growing interest in the self-correction capability of Large Language Models (LLMs), there are varying conclusions about its effectiveness. Prior research has largely concentrated on intrinsic self-correction, extrinsic self-correction, particularly the interplay between internal knowledge and external feedback, remains underexplored. In this paper, we aim to comprehensively investigate the underlying mechanism of moral self-correction by addressing a fundamental question: is moral self-correction an innate capability of LLMs? Specifically, we conduct: (1) a behavioral analysis of LLMs' moral sensitivity based on a self-distinguishing task; and (2) a mechanistic analysis of the hidden states to examine how key components of self-correction, such as Chain-of-Thought (CoT) and external feedback, interact to facilitate moral self-correction. Drawing on empirical evidence from both behavioral and mechanistic analyses, we demonstrate that moral self-correction is not an inherent capability of LLMs, as they are neither morally sensitive nor able to effectively incorporate external feedback during the self-correction process.
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。
