AI生成图像严重偏见，女性、黑人、残障者被边缘化
Generative AI Carries Non-Democratic Biases and Stereotypes: Representation of Women, Black Individuals, Age Groups, and People with Disability in AI-Generated Images across Occupations
- 分析37职业444张AI生成图，发现群体代表性严重失衡
- 黑人几乎缺席，残障人士全无，女性在技术岗占比极低
- 适合关注AI伦理、数字公平的开发者与政策制定者阅读
本研究探讨生成式人工智能系统如何复制并强化社会偏见，重点关注女性、黑人群体、年龄组及可见残障人士在AI生成的职业图像中的表现。分析了微软Designer、Meta AI和Ideogram生成的444张图像，覆盖37个职业领域。结果显示：女性在高层及科技岗位中严重缺位，黑人群体近乎完全缺席，所有类别中均未出现可见残障人士；同时存在明显年龄偏向，多为年轻形象。这些模式表明，生成式AI工具再现甚至放大了职场不平等与刻板印象，损害了公平与包容的民主价值。研究呼吁开发者与企业用户对AI工具进行公平性、多样性与包容性（EDI）风险审计，并强调应让多元群体参与AI开发与治理，以推动更具社会责任感的技术。
原文摘要 · Abstract (English)
In this study, I investigate how generative artificial intelligence (AI) systems reproduce and reinforce societal biases, with a specific focus on the representation of women, Black individuals, age groups, and people with visible disabilities in AI-generated occupational images. I analyzed 444 images generated by Microsoft Designer, Meta AI, and Ideogram across 37 occupations and found significant disparities in representation. Women are underrepresented in senior and technology roles, Black individuals are nearly absent, and people with visible disabilities are completely absent across all categories. I also observed clear age bias, with younger individuals predominantly depicted. These patterns suggest that generative AI tools replicate, and in some cases amplify, existing workplace inequalities and stereotypes, undermining democratic values of equity and inclusion. My findings highlight the urgent need for algorithmic diversity exposure, and I recommend that AI developers and corporate users audit their tools for equity, diversity, and inclusion (EDI) risks. I argue for the critical inclusion of diverse groups in AI development and governance to foster more democratic and socially responsible technologies.
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。