测试7大主流大模型对仇恨言论的反应，发现其处理方式差异显著。
Decoding Hate: Exploring Language Models' Reactions to Hate Speech
- 对比分析7个大模型在面对仇恨言论时的响应策略。
- 部分模型能识别并拒绝生成仇恨内容，但效果不一。
- 适合关注AI伦理与安全的开发者和研究者阅读。
仇恨言论是在线环境中一种有害的表达形式，常表现为侮辱性内容，构成重大数字风险。随着大型语言模型（LLMs）的兴起，人们对其可能复制网络数据中的仇恨言论模式表示担忧，因其训练数据包含大量未经审核的内容。理解这些模型对仇恨言论的反应对于其负责任的应用至关重要。然而，现有研究对此关注有限。本文探讨了七种前沿大模型（LLaMA 2、Vicuna、LLaMA 3、Mistral、GPT-3.5、GPT-4 和 Gemini Pro）对仇恨言论的反应。通过定性分析，揭示了这些模型在应对仇恨言论输入时的多样化表现，突显其处理能力的差异。同时讨论了通过微调和提示引导等策略减轻仇恨言论生成的方法。最后，研究还考察了模型对以政治正确语言包装的仇恨言论的回应。
原文摘要 · Abstract (English)
Hate speech is a harmful form of online expression, often manifesting as derogatory posts. It is a significant risk in digital environments. With the rise of Large Language Models (LLMs), there is concern about their potential to replicate hate speech patterns, given their training on vast amounts of unmoderated internet data. Understanding how LLMs respond to hate speech is crucial for their responsible deployment. However, the behaviour of LLMs towards hate speech has been limited compared. This paper investigates the reactions of seven state-of-the-art LLMs (LLaMA 2, Vicuna, LLaMA 3, Mistral, GPT-3.5, GPT-4, and Gemini Pro) to hate speech. Through qualitative analysis, we aim to reveal the spectrum of responses these models produce, highlighting their capacity to handle hate speech inputs. We also discuss strategies to mitigate hate speech generation by LLMs, particularly through fine-tuning and guideline guardrailing. Finally, we explore the models' responses to hate speech framed in politically correct language.
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。