文本到图像系统通过修改提示词引入文化偏见
Prompt Revision as a Source of Cultural Bias in Text-to-Image Systems
朋友间推荐：最近发现一个挺有意思的研究，他们分析了 DALL-E-3、Imagen-4 这些大模型，发现它们在生成图片前会偷偷修改你的提示词，而且这个修改过程本身就会引入文化偏见，比如把非西方语境简化成刻板印象，挺有意思的。
研究团队发现，DALL-E-3、Imagen-4 和 GPT-Image-1.5 等商业文本到图像系统在生成图像前会修改用户提示词，这一步骤通常不可见。他们使用 WORLDVIEW 基准，包含 8960 个多语言提示词，审计了三个系统的修改层。研究发现，与无语境的英文基准相比，美国语境被标记得最少，而非西方和非英语语境则被标记得更重，并简化为狭窄的词汇，这些词汇被应用于不同主题的提示词，最终形成可识别的文化刻板印象。研究指出，这种刻板印象的来源是系统本身，而非模型本身。
Prompt Revision as a Source of Cultural Bias in Text-to-Image Systems
Commercial text-to-image systems silently revise user prompts before generating images, a step users typically cannot disable or even see. Yet, existing audits of cultural bias examine only the final images and treat generation as a single pipeline, so they cannot tell where the bias originates. We introduce WORLDVIEW, a multilingual benchmark of 8,960 prompts across 15 languages and 31 language-context pairings. Using it, we audit the revision layer in three systems (DALL-E-3, Imagen-4, GPT-Image-1.5) through a three-step analysis of how heavily it marks each cultural context, whether it flattens that context into a narrow vocabulary, and whether that vocabulary is stereotypical. Relative to a no-context English baseline, the US is the least-marked context, while non-Western and non-Anglophone contexts are marked far more heavily, flattened into narrow vocabularies applied across topically diverse prompts, and reduced to recognizable cultural stereotypes. Comparing images from original versus revised prompts on models without a revision layer, we identify the layer itself as a previously undocumented, causal source of this stereotyping. To locate cultural bias, and fix it, we must audit the system as deployed, not the model alone.