PulseAugur
实时 10:49:13
English(EN) Who Brought Easter Eggs to Eid? Auditing Cultural Translation of Math Word Problems Across Diverse Languages and Regions

大型语言模型在数学应用题的文化翻译方面存在困难

一项新研究分析了像Claude Opus 4、GPT-4.1和Gemini 2.5 Pro这样的大型语言模型如何跨越不同语言和文化翻译数学应用题。研究发现,尽管模型通常在转换类型上达成一致,但它们经常替换特定的文化元素,如姓名和食物,导致呈现给学生的文化背景产生显著差异。此外,所有测试的语言-模型组合都表现出“熵坍缩”,这意味着适应过程压缩而非扩展了文化多样性,模型经常错误地归因于区域背景或引入跨文化污染,例如将寻蛋活动等同于开斋节活动。 AI

影响 揭示了大型语言模型在细致的文化翻译能力方面存在显著局限性,影响了教育应用。

排序理由 该集群包含一篇详细介绍大型语言模型能力研究结果的学术论文。

在 arXiv cs.CL 阅读 →

AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →

报道来源 [2]

  1. arXiv cs.CL TIER_1 English(EN) · Parisa Suchdev, Juniper Lovato ·

    Who Brought Easter Eggs to Eid? Auditing Cultural Translation of Math Word Problems Across Diverse Languages and Regions

    arXiv:2606.11009v1 Announce Type: new Abstract: Large language models are increasingly used to adapt math word problems for personalized learning at scale, but it remains an open question whether those adaptations are consistent across models, preserve cultural diversity at scale…

  2. arXiv cs.CL TIER_1 English(EN) · Juniper Lovato ·

    Who Brought Easter Eggs to Eid? Auditing Cultural Translation of Math Word Problems Across Diverse Languages and Regions

    Large language models are increasingly used to adapt math word problems for personalized learning at scale, but it remains an open question whether those adaptations are consistent across models, preserve cultural diversity at scale, and reveal which cultural entities models trea…