Amplification of conventional female gender displays and reinforcement of Asian stereotypes in user-generated Artificial Intelligence (AI) images on social media

Abstract

In a content analysis of user-generated artificial intelligence (AI) images from a public Facebook group, the authors investigate how female representations reflect normative racial and gendered stereotypes. Their results suggest that the proportion of conventional feminine expressions, as determined by Erving Goffman’s typology of gender display in Gender Advertisements (1979), are similar to late 20th to early 21st-century magazine advertisements. Multiple degrees of gender display in AI media point to exaggerations of gender display that could be a concern for future AI tool training. Levels of sexualization were high and similar across racial groups. Asians were more likely to be represented in ways consistent with the geisha stereotype and with a pinup motif that reinforces female subordination.