“AI Cannot Speak African”: An Intersectional Feminist Perspective on South African Women's AI Engagement

ABSTRACT Using an intersectional feminist perspective, this study investigates intersectional inequalities in artificial intelligence (AI) engagement among underrepresented non‐Western women. The study employs a qualitative approach and interviews 16 corporate women in South Africa of different races. The findings reveal the compounding impact of intersectional marginalities, namely, gender, race, language, and non‐Western geographical location. The deficit narratives of Africa and African women in the AI‐generated outputs, lack of inclusive design in the AI systems, and weak governance in South Africa further produced exclusions in broader knowledge infrastructures and historically constituted geopolitical relations of technological power. This study contributes to the literature by revealing that the current AI landscape is perpetuating not only the exclusion of women but also that of non‐Western and specifically African sources. South Africa is not only an empirical site to examine feminist AI concerns but also offers a unique context to theorize how AI reproduces gendered hierarchies alongside wider geopolitical hierarchies.