arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2312.09085cs.CLcs.AIcs.CRcs.CY
收藏

地球是平的因为……:通过说服性对话调查 LLM 对错误信息的信念

The Earth is Flat because...: Investigating LLMs' Belief towards Misinformation via Persuasive Conversation

Tsinghua University(清华大学);Nanyang Technological University(南洋理工大学);Shanghai Jiao Tong University(上海交通大学);Stanford University(斯坦福大学)

Rongwu Xu, Brian S. Lin, Shujian Yang, Tianqi Zhang, Weiyan Shi, Tianwei Zhang, Zhixuan Fang, Wei Xu, Han Qiu

AI总结:

本研究通过构建 Farm 数据集和测试框架,深入探讨了大型语言模型在多轮说服性对话中对错误信息的易感性,发现其正确的事实信念极易被说服策略操纵。

摘要:

大型语言模型(LLMs)掌握了大量知识,但仍然容易受到外部错误信息的影响。现有研究主要在单轮设置下研究这种易受影响的行为。然而,信念在多轮对话中可能会发生改变,尤其是在说服性对话中。因此,在本研究中,我们深入探讨了 LLMs 对说服性对话的易感性,特别是针对它们能够正确回答的事实性问题。我们首先策划了 Farm(即 Fact to Misinform)数据集,其中包含事实性问题以及系统生成的说服性错误信息。然后,我们开发了一个测试框架来追踪 LLMs 在说服性对话中的信念变化。通过大量实验,我们发现 LLMs 对事实知识的正确信念很容易被各种说服策略所操纵。

英文摘要:

Large language models (LLMs) encapsulate vast amounts of knowledge but still remain vulnerable to external misinformation. Existing research mainly studied this susceptibility behavior in a single-turn setting. However, belief can change during a multi-turn conversation, especially a persuasive one. Therefore, in this study, we delve into LLMs' susceptibility to persuasive conversations, particularly on factual questions that they can answer correctly. We first curate the Farm (i.e., Fact to Misinform) dataset, which contains factual questions paired with systematically generated persuasive misinformation. Then, we develop a testing framework to track LLMs' belief changes in a persuasive dialogue. Through extensive experiments, we find that LLMs' correct beliefs on factual knowledge can be easily manipulated by various persuasive strategies.

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。

↑