LLM 是有偏见的教师：评估个性化教育中的 LLM 偏见
LLMs are Biased Teachers: Evaluating LLM Bias in Personalized Education
AI总结：
本文评估了 LLM 在个性化教育中作为“教师”时的偏见，提出 MAB 和 MDB 两种指标，基于 9 个模型和超 17,000 条教育解释发现模型会延续或反转刻板印象，且收入与残障状况相关偏见最突出。
摘要：
随着大型语言模型（LLMs）在教育领域的日益普及，对这些模型固有偏见的担忧也日益突出。我们在个性化教育环境中评估 LLM 的偏见，特别关注这些模型作为“教师”的角色。我们揭示了模型在生成和选择面向不同人口群体的教育内容时存在的显著偏见，这些群体包括种族、民族、生理性别、社会性别、残障状况、收入和国籍。我们引入并应用了两种偏见评分指标——平均绝对偏见（MAB）和最大差异偏见（MDB）——来分析 9 个开源和闭源的最先进 LLM。我们的实验利用了超过 17,000 条跨多个难度级别和主题的教育解释，发现模型可能通过延续有害刻板印象和反转这些刻板印象来损害学生学习。我们发现所有前沿模型的偏见程度相似，其中 MAB 在收入水平上最高，而 MDB 在收入和残障状况两个维度上均最高。对于这两种指标，我们发现生理性别/社会性别以及种族/民族的偏见最低。
英文摘要：
With the increasing adoption of large language models (LLMs) in education, concerns about inherent biases in these models have gained prominence. We evaluate LLMs for bias in the personalized educational setting, specifically focusing on the models' roles as "teachers." We reveal significant biases in how models generate and select educational content tailored to different demographic groups, including race, ethnicity, sex, gender, disability status, income, and national origin. We introduce and apply two bias score metrics--Mean Absolute Bias (MAB) and Maximum Difference Bias (MDB)--to analyze 9 open and closed state-of-the-art LLMs. Our experiments, which utilize over 17,000 educational explanations across multiple difficulty levels and topics, uncover that models potentially harm student learning by both perpetuating harmful stereotypes and reversing them. We find that bias is similar for all frontier models, with the highest MAB along income levels while MDB is highest relative to both income and disability status. For both metrics, we find the lowest bias exists for sex/gender and race/ethnicity.
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。
