"Suits as Masculine and Flowers as Feminine" : Investigating Gender Expression in AI-Generated Imagery

Collyer-Hoar, Gail and Rubegni, Elisa and Tomczyk, Ben and Baines, Alexander and Gruia, Lidia (2025) "Suits as Masculine and Flowers as Feminine" : Investigating Gender Expression in AI-Generated Imagery. In: DIS '25: Proceedings of the 2025 ACM Designing Interactive Systems Conference :. ACM, New York, pp. 915-928. ISBN 9798400714856

Text (Suits_as_Masculine_and_Flowers_as_Feminine____Investigating_Gender_Expression_in_AI_Generated_Imagery)
Suits_as_Masculine_and_Flowers_as_Feminine_Investigating_Gender_Expression_in_AI_Generated_Imagery.pdf - Accepted Version
Available under License Creative Commons Attribution.
Download (3MB)

Abstract

Generative AI’s growing use in content creation significantly impacts societal perceptions by perpetuating and reinforcing gender stereotypes. The amplification of stereotypes in AI-generated content can lead to increased discrimination, exclusion, misinformation, and contribute to racial and gender disparities. To address this challenge, we explore the direct impact of generative AI on gender attribution and stereotype reinforcement in digital imagery through a survey with 111 participants, analysing interpretations of gender expression in 216 AI-generated images. Findings reveal a pronounced bias toward masculine-leaning attributions, particularly in images where gender identity is labelled as androgyne. This research provides three key contributions: (1) an in-depth understanding of how people perceive gender expressions in AI-generated images; (2) a dataset of 216 images evaluated by participants for masculinity, femininity, and neutrality; and (3) two key challenges to consider in order to address the stereotyped representations of gender expressions in AI-generated content, highlighting the need for more inclusive AI practices.