用提示词控制生成图像中职业形象的性别种族分布，发现效果因模型而异。
Prompting Away Stereotypes? Evaluating Bias in Text-to-Image Models for Occupations
- 设计公平提示词对比中性提示，评估模型对职业形象的偏见
- 不同模型响应差异大：有的显著改善多样性，有的过度均匀化
- 提示词有效但有限，需结合模型级改进策略
文本生成图像（TTI）模型虽具强大创作能力，但可能加剧有害社会偏见。我们将表征性社会偏见评估视为图像筛选与评价任务，构建首个涵盖五类社会敏感职业（首席执行官、护士、软件工程师、教师、运动员）的基准数据集。选用五种先进模型：闭源（DALL·E 3、Gemini Imagen 4.0）与开源（FLUX.1-dev、Stable Diffusion XL Turbo、Grok-2 Image），比较中性基线提示与公平导向控制提示的效果。所有生成图像均标注性别（男/女）和种族（亚裔、黑人、白人），支持分布分析。结果显示，提示词能显著改变性别种族分布，但效果高度依赖模型：部分系统有效提升多样性，部分过度纠正导致不真实均匀化，部分几乎无响应。这揭示了提示词作为公平干预手段的潜力与局限，强调需配合模型层面策略。代码与数据已公开：https://github.com/maximus-powers/img-gen-bias-analysis。
原文摘要 · Abstract (English)
Text-to-Image (TTI) models are powerful creative tools but risk amplifying harmful social biases. We frame representational societal bias assessment as an image curation and evaluation task and introduce a pilot benchmark of occupational portrayals spanning five socially salient roles (CEO, Nurse, Software Engineer, Teacher, Athlete). Using five state-of-the-art models: closed-source (DALLE 3, Gemini Imagen 4.0) and open-source (FLUX.1-dev, Stable Diffusion XL Turbo, Grok-2 Image), we compare neutral baseline prompts against fairness-aware controlled prompts designed to encourage demographic diversity. All outputs are annotated for gender (male, female) and race (Asian, Black, White), enabling structured distributional analysis. Results show that prompting can substantially shift demographic representations, but with highly model-specific effects: some systems diversify effectively, others overcorrect into unrealistic uniformity, and some show little responsiveness. These findings highlight both the promise and the limitations of prompting as a fairness intervention, underscoring the need for complementary model-level strategies. We release all code and data for transparency and reproducibility https://github.com/maximus-powers/img-gen-bias-analysis.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。