用新方法发现Sora生成视频隐含种族性别偏见，且去偏提示可能适得其反。
VEAT Quantifies Implicit Associations in Text-to-Video Generator Sora and Reveals Challenges in Bias Mitigation
- 提出VEAT和SC-VEAT测试视频中隐性关联，扩展了文本图像评估范式。
- 欧洲裔与女性在17个职业中更关联愉悦感（效应量d>0.8），与真实人口分布高度相关。
- 去偏提示反而强化部分职业的种族刻板印象，警示模型部署风险。
文本到视频生成模型如Sora引发了对生成内容是否反映社会偏见的担忧。本文将嵌入关联测试从词与图像扩展至视频，提出视频嵌入关联测试（VEAT）与单类别VEAT（SC-VEAT）。通过复现主流基线中的关联方向与强度，包括隐式联想测试（IAT）场景与OASIS图像类别，验证了方法有效性。进一步量化了17个职业与7项奖项中，非洲裔与欧洲裔、女性与男性在愉悦性（愉悦 vs. 不悦）上的隐性关联。结果表明，欧洲裔与女性在多数情境中更关联愉悦感（效应量均大于0.8）。效应量与现实人口分布高度相关：职业中男性与白人占比（r=0.93, r=0.83），奖项中男性与非黑人占比（r=0.88, r=0.99）。使用显式去偏提示通常降低效应量，但存在反效果：两名与黑人相关的职位（清洁工、邮政服务）经去偏后反而更强化黑人关联。这些结果揭示，若未经严格评估与负责任部署，通用文本到视频生成器可能放大表征伤害。
原文摘要 · Abstract (English)
Text-to-Video (T2V) generators such as Sora raise concerns about whether generated content reflects societal bias. We extend embedding-association tests from words and images to video by introducing the Video Embedding Association Test (VEAT) and Single-Category VEAT (SC-VEAT). We validate these methods by reproducing the direction and magnitude of associations from widely used baselines, including Implicit Association Test (IAT) scenarios and OASIS image categories. We then quantify race (African American vs. European American) and gender (women vs. men) associations with valence (pleasant vs. unpleasant) across 17 occupations and 7 awards. Sora videos associate European Americans and women more with pleasantness (both d>0.8). Effect sizes correlate with real-world demographic distributions: percent men and White in occupations (r=0.93, r=0.83) and percent male and non-Black among award recipients (r=0.88, r=0.99). Applying explicit debiasing prompts generally reduces effect-size magnitudes, but can backfire: two Black-associated occupations (janitor, postal service) become more Black-associated after debiasing. Together, these results reveal that easily accessible T2V generators can actually amplify representational harms if not rigorously evaluated and responsibly deployed.
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。