AI 中文总结
该研究提出MECSS框架测量大语言模型的东方主义结构性话语偏见，发现GPT-4和Falcon3-7B-Instruct均存在系统性东方主义模式，且Falcon3-7B-Instruct得分更高，Said-washing现象普遍存在。
AI 中文摘要
人工智能系统如今塑造着数亿人了解其他文化的方式。当有人向这些系统询问中东相关问题时，得到的并非中立事实，而是由训练数据中嵌入的框架所塑造的呈现形式，而这些数据绝大多数是西方的英文数据。本文探究这种呈现形式是否符合萨义德所定义的东方主义：是否否认中东行为体的能动性、将西方框架视为中立而把非西方知识标记为特殊性、通过自身未产生的范畴来解释该地区。标准公平性指标无法回答这一问题，因为它们检测的是明确的偏见而非结构性框架。本文提出中东文化敏感性评分（Middle East Cultural Sensitivity Score，MECSS）框架，将萨义德的七种东方主义操作转化为可测量的维度，并提出“Said-washing”一词指代一种特定失败：模型声称不进行泛化，却复制了其否认的结构。在280次对话（1120轮交互）中，GPT-4和Falcon3-7B-Instruct均系统性地复制东方主义模式，且通过结构性定位而非公开刻板印象实现。GPT-4得分中等（平均MECSS为1.73）；Falcon3-7B-Instruct得分更高（2.18），尽管它在阿布扎比构建并使用阿拉伯语内容训练。这反驳了“在区域内构建模型会降低其东方主义倾向”的假设，不过这些模型在规模和来源上存在差异，因此无法将地理因素单独分离为原因。认知中心（将西方框架视为无标记的普遍性）在两个模型的评分中均接近顶端。Said-washing出现在87.9%的GPT-4对话中，这是现有指标无法察觉的模式。减少这种偏见需要改变模型的学习内容，而非仅增加语言或迁移机构位置。
英文摘要
AI systems now shape how hundreds of millions of people learn about cultures other than their own. When someone asks one of these systems about the Middle East, they do not receive neutral facts. They receive a representation shaped by the frameworks embedded in training data, and that data is overwhelmingly Western and English-language. This paper asks whether that representation is Orientalist in Said's sense: whether it denies agency to Middle Eastern actors, treats Western frameworks as neutral while marking non-Western knowledge as particular, and explains the region through categories it did not produce. Standard fairness metrics cannot answer this, because they detect explicit prejudice rather than structural framing. This paper introduces the Middle East Cultural Sensitivity Score (MECSS), a framework that turns Said's seven Orientalist operations into measurable dimensions, and the term "Said-washing" for a specific failure: a model that disclaims generalization, then reproduces the structure it disclaimed. Across 280 conversations (1,120 exchanges), GPT-4 and Falcon3-7B-Instruct both reproduce Orientalist patterns systematically, through structural positioning rather than open stereotyping. GPT-4 scores moderately (mean MECSS 1.73); Falcon3-7B-Instruct scores higher (2.18), even though it was built in Abu Dhabi and trained with Arabic content. This is evidence against the assumption that building a model regionally makes it less Orientalist, though the models differ in size as well as origin, so geography cannot be isolated as the cause. Epistemic Center, the treatment of Western frameworks as unmarked universals, scores near the top of the scale for both models. Said-washing appears in 87.9% of GPT-4 conversations, a pattern existing metrics cannot see. Reducing this bias requires changing what models learn from, not only adding languages or relocating institutions.
Comments16 pages, 3 tables
