Differentiated or Degrading? Exploring Users’ Perceptions and Potential Implications of Gender Bias in Chatbot Interactions

Generative AI systems, such as ChatGPT, have become increasingly integrated into everyday life. However, the issue of gender bias in AI has garnered increasing public and scholarly attention. This study thus explores users’ perceptions and potential implications of gender bias in generative AI, including the role of certain user predispositions and chatbot affordances. A qualitative, stimulus-based user study combining chatbot interactions and semi-structured interviews revealed a diverse range of perceptions among users, as well as nuanced perspectives within participants. These perceptions appear determined by users’ own gender role attitudes, their enthusiasm for AI, as well as their expectations of neutrality toward AI. Further, the chatbot affordances human-likeness, personalization, and conversation control seem to shape users’ opinions. Moreover, while identifying selective usage, shifts in source perception, and the consolidation of gender roles as potential outcomes, our findings also point to users’ limited reflection on possible implications of gender bias in AI.