Framed by Gender: Users’ Interactions with Generative Artificial Intelligence

• In Event: 30611 - Regular Session on Gender: Femininities and Masculinities

Abstract

Gender scholars argue that gender works as a primary cultural frame during interpersonal
interactions, with gender categorization and stereotyping extending into innovative contexts. The
emergence of generative artificial intelligence (GAI) systems, such as ChatGPT, marks a new era
of human-machine interaction. In this article, we ask whether and how users apply the gender
frame when interacting with these new GAI systems. We gathered a corpus of 0.86 million
online Twitter posts mentioning ChatGPT over a two-month period in early 2023, and examine
how users assign a gender-binary pronoun to ChatGPT. A preliminary computational text
analysis of 10,000 sampled posts reveals that 12 percent contained gendered pronouns to refer to
ChatGPT, with masculine pronouns (69 percent) predominating over feminine ones (31 percent).
Results from topic models further suggest that the constructed masculinity in ChatGPT is
associated with its perceived general intelligence and remarkable capacities in a diverse range of
tasks. The feminine traits, in contrast, are manifested in scenarios when ChatGPT did not achieve
expected performance or excelled at stereotypically feminine tasks. These findings demonstrate
the process of gender framing in emerging forms of human-GAI interaction, which perpetuates
traditional and harmful gender stereotypes that reproduces gender inequality in everyday
interactions cumulatively .

Authors

• Man Yao, Denison University

• Claudia Buchmann, Ohio State University

• Zhen Wang, UC San Diego