arXiv 2511.172562025-11-21cs.CYcs.CL

跨文化价值观对齐框架用于负责任的AI治理：来自中西方比较分析的证据

Cross-cultural value alignment frameworks for responsible AI governance: Evidence from China-West comparative analysis

Wuhan University of Science and Technology（武汉理工大学）；The Hong Kong University of Science and Technology (Guangzhou)（香港科学与技术大学（广州））；University of Copenhagen（哥本哈根大学）；WUST-Madrid Complutense Institute（武汉理工大学-马德里康普顿斯研究所）

Haijiang Liu, Jinguang Gu, Xun Wu, Daniel Hershcovich, Qiaoling Xiao

备注：: Presented on Academic Conference "Technology for Good: Driving Social Impact" (2025)

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

本研究提出多层审计平台，评估中西方LLM跨文化价值观对齐，发现中国模型在多语言优化上更优，而西方模型存在美国偏见，Mistral系列在跨文化对齐上表现突出。

摘要：

随着大型语言模型（LLMs）越来越多地影响全球范围内的高风险决策，确保其与不同文化价值观的一致性已成为一个关键的治理挑战。本研究提出了一种多层审计平台，用于负责任的AI，通过四种整合的方法系统地评估中国起源和西方起源的LLMs中的跨文化价值观对齐：伦理困境语料库用于评估时间稳定性，多样性增强框架（DEF）用于量化文化保真度，第一令牌概率对齐用于分布准确性，以及多阶段推理框架（MARK）用于可解释的决策制定。我们对20多个领先模型的比较分析，如Qwen、GPT-4o、Claude、LLaMA和DeepSeek，揭示了普遍的挑战——价值观系统的根本不稳定性、年轻人口的系统性代表性不足以及模型规模与对齐质量之间的非线性关系，以及不同的地区发展轨迹。虽然中国起源的模型越来越强调多语言数据整合以实现特定情境的优化，西方模型则表现出更大的架构实验性，但仍然存在以美国为中心的偏见。没有一种范式能够实现稳健的跨文化泛化。我们发现Mistral系列架构在跨文化对齐方面显著优于LLaMA3系列，而全参数微调在多样化数据集上的表现优于强化学习从人类反馈中学习，在保持文化多样性方面更优。

英文摘要：

As Large Language Models (LLMs) increasingly influence high-stakes decision-making across global contexts, ensuring their alignment with diverse cultural values has become a critical governance challenge. This study presents a Multi-Layered Auditing Platform for Responsible AI that systematically evaluates cross-cultural value alignment in China-origin and Western-origin LLMs through four integrated methodologies: Ethical Dilemma Corpus for assessing temporal stability, Diversity-Enhanced Framework (DEF) for quantifying cultural fidelity, First-Token Probability Alignment for distributional accuracy, and Multi-stAge Reasoning frameworK (MARK) for interpretable decision-making. Our comparative analysis of 20+ leading models, such as Qwen, GPT-4o, Claude, LLaMA, and DeepSeek, reveals universal challenges-fundamental instability in value systems, systematic under-representation of younger demographics, and non-linear relationships between model scale and alignment quality-alongside divergent regional development trajectories. While China-origin models increasingly emphasize multilingual data integration for context-specific optimization, Western models demonstrate greater architectural experimentation but persistent U.S.-centric biases. Neither paradigm achieves robust cross-cultural generalization. We establish that Mistral-series architectures significantly outperform LLaMA3-series in cross-cultural alignment, and that Full-Parameter Fine-Tuning on diverse datasets surpasses Reinforcement Learning from Human Feedback in preserving cultural variation...

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。