This study examines the manifestations of sexism, racism, and classism in the output of six text-to-image generative AI systems within the constructs of power, success, and beauty. A total of 180 images were generated using three prompts for each AI tool. Our analysis focused on detecting gender, racial, and class biases, as well as age discrimination. The findings reveal an underrepresentation of women and People of Color across the generated images. Additionally, the tendency to depict women in a sexualized manner was prominent. Data also indicated a bias towards younger depictions of women relative to men and People of Color relative to white individuals. The images overwhelmingly represented individuals as belonging to a higher socioeconomic class, pointing towards a systemic bias within AI systems towards privilege.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Sexism, Racism, and Classism: Social Biases in Text-to-Image Generative AI in the Context of Power, Success, and Beauty

  • Eva Johanna Gengler

摘要

This study examines the manifestations of sexism, racism, and classism in the output of six text-to-image generative AI systems within the constructs of power, success, and beauty. A total of 180 images were generated using three prompts for each AI tool. Our analysis focused on detecting gender, racial, and class biases, as well as age discrimination. The findings reveal an underrepresentation of women and People of Color across the generated images. Additionally, the tendency to depict women in a sexualized manner was prominent. Data also indicated a bias towards younger depictions of women relative to men and People of Color relative to white individuals. The images overwhelmingly represented individuals as belonging to a higher socioeconomic class, pointing towards a systemic bias within AI systems towards privilege.