When AI is sexist: investigating responses to AI and human discrimination

Official URL: https://dx.doi.org/10.20368/1971-8829/1136320

As artificial intelligence (AI) systems prove their utility in various aspects of daily life, their potential risk to societal progress is still debated given the increasing rates of their discriminatory behaviors. Some preliminary evidence suggests that the same discriminatory behavior may be perceived differently when performed by a person or an AI system, yet it remains largely unknown how people perceive and respond to AI’s discriminatory behaviors. Through two studies conducted in Türkiye (Study 1) and the United States (Study 2), participants read about either AI’s discrimination against women, humans’ discrimination against women, or an unrelated topic, and indicated the degree of perceived threat to women, willingness to participate in collective action for women’s equal treatment, and their attitudes toward women and feminists. In both cultures, discrimination was perceived to be less threatening to women when committed by AI compared to a human, yet collective action intentions and attitudes toward women and feminists did not differ across conditions. Notably, American men consistently reported less favorable attitudes toward women after reading about discrimination. These findings suggest discrimination by AI requires special attention and underline the importance of investigating the influence of technologies on sustaining intergroup inequality.