迈向包容性教育AI：通过多元性视角审计前沿大语言模型
Toward Inclusive Educational AI: Auditing Frontier LLMs through a Multiplexity Lens
AI总结：
本文针对教育场景中前沿 LLM 的 WEIRD 文化偏见，提出多元性审计框架，并比较系统提示嵌入与多智能体协商两种缓解策略，证明 MAS 可将 PDS 熵由 3.25% 提升至 98%。
摘要：
随着 GPT-4 和 Llama 3 等大语言模型（LLM）成为教育场景中不可或缺的组成部分，人们对这些技术内嵌的文化偏见、权力失衡和伦理局限的担忧正不断加剧。尽管生成式 AI 工具旨在改善学习体验，但它们往往反映植根于西方、受过教育、工业化、富裕和民主（WEIRD）文化范式的价值观，可能使多元的全球视角被边缘化。本文提出一个通过应用多元性（multiplexity）视角评估并缓解 LLM 中文化偏见的框架。多元性受 Senturk 等人启发，并植根于伊斯兰及其他智慧传统，强调多元文化观点的共存，支持一种整合经验科学与规范性价值的多层认识论。我们的分析表明，LLM 经常表现出文化两极化，偏见既出现在显性回答中，也出现在微妙的语境线索中。为应对固有偏见并将多元性纳入 LLM，我们提出两种策略：Contextually-Implemented Multiplex LLMs（语境实现的多元 LLM），将多元原则直接嵌入系统提示，在基础层面影响 LLM 输出，并且独立于单个提示；以及 Multi-Agent System (MAS)-Implemented Multiplex LLMs（多智能体系统实现的多元 LLM），其中多个 LLM 智能体各自代表不同文化观点，协作生成平衡、综合的回答。研究结果表明，随着缓解策略从语境提示发展到 MAS 实现，文化包容性显著提升，其证据是 Perspectives Distribution Score（PDS，视角分布分数）显著上升，且 PDS 熵从基线的 3.25% 增至采用 MAS-Implemented Multiplex LLMs 时的 98%。情感分析进一步显示，跨文化情感出现向积极情感的转变，……
英文摘要：
As large language models (LLMs) like GPT-4 and Llama 3 become integral to educational contexts, concerns are mounting over the cultural biases, power imbalances, and ethical limitations embedded within these technologies. Though generative AI tools aim to enhance learning experiences, they often reflect values rooted in Western, Educated, Industrialized, Rich, and Democratic (WEIRD) cultural paradigms, potentially sidelining diverse global perspectives. This paper proposes a framework to assess and mitigate cultural bias within LLMs through the lens of applied multiplexity. Multiplexity, inspired by Senturk et al. and rooted in Islamic and other wisdom traditions, emphasizes the coexistence of diverse cultural viewpoints, supporting a multi-layered epistemology that integrates both empirical sciences and normative values. Our analysis reveals that LLMs frequently exhibit cultural polarization, with biases appearing in both overt responses and subtle contextual cues. To address inherent biases and incorporate multiplexity in LLMs, we propose two strategies: \textit{Contextually-Implemented Multiplex LLMs}, which embed multiplex principles directly into the system prompt, influencing LLM outputs at a foundational level and independent of individual prompts, and \textit{Multi-Agent System (MAS)-Implemented Multiplex LLMs}, where multiple LLM agents, each representing distinct cultural viewpoints, collaboratively generate a balanced, synthesized response. Our findings demonstrate that as mitigation strategies evolve from contextual prompting to MAS-implementation, cultural inclusivity markedly improves, evidenced by a significant rise in the Perspectives Distribution Score (PDS) and a PDS Entropy increase from 3.25\% at baseline to 98\% with the MAS-Implemented Multiplex LLMs. Sentiment analysis further shows a shift towards positive sentiment across cultures,...
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。
