“既然律师都是男性……”:审查 LLM 印地语生成中的隐性性别偏见
'Since Lawyers are Males..': Examining Implicit Gender Bias in Hindi Language Generation by LLMs
AI总结:
本研究探讨了 LLM 在印地语文本生成中的隐性性别偏见,开发了受 WinoBias 启发的印地语数据集测试 GPT-4o 和 Claude-3 sonnet,发现印地语性别偏见高达 87.8%,远超英语的 33.4%,揭示了跨语言偏见差异及应对考量。
摘要:
大型语言模型(LLMs)正越来越多地被用于生成各种语言的文本,应用于翻译、客户支持和教育等任务。尽管取得了这些进展,LLMs 在英语中表现出显著的性别偏见,而在生成印地语等代表性相对不足的语言内容时,这种偏见变得更加明显。本研究探讨了印地语文本生成中的隐性性别偏见,并将其与英语中的偏见进行比较。我们开发了受 WinoBias 启发的印地语数据集,以检查 GPT-4o 和 Claude-3 sonnet 等模型响应中的刻板模式。结果显示,印地语中存在 87.8% 的显著性别偏见,而英语 GPT-4o 生成中为 33.4%,印地语响应频繁依赖与职业、权力等级和社会阶层相关的性别刻板印象。本研究强调了不同语言间性别偏见的差异,并为在生成式 AI 系统中应对这些偏见提供了参考。
英文摘要:
Large Language Models (LLMs) are increasingly being used to generate text across various languages, for tasks such as translation, customer support, and education. Despite these advancements, LLMs show notable gender biases in English, which become even more pronounced when generating content in relatively underrepresented languages like Hindi. This study explores implicit gender biases in Hindi text generation and compares them to those in English. We developed Hindi datasets inspired by WinoBias to examine stereotypical patterns in responses from models like GPT-4o and Claude-3 sonnet. Our results reveal a significant gender bias of 87.8% in Hindi, compared to 33.4% in English GPT-4o generation, with Hindi responses frequently relying on gender stereotypes related to occupations, power hierarchies, and social class. This research underscores the variation in gender biases across languages and provides considerations for navigating these biases in generative AI systems.
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。