Reproduction of social biases through AI: a study on AI developers' awareness on social biases

Artificial Intelligence (AI) stands as an evolving and controversial force, changing the way of work in numerous sectors including education, art, finance, health, transportation, and security, also having influences over daily lives. While discussions often revolve around the technological breakthroughs and economic ramifications of AI, it is imperative to recognize and address the social impacts and consequences it carries, particularly with respect to potential social biases and discriminations that it will (re)produce or contribute to. This study explores AI developers' awareness and perspectives on gender, race and ethnicity-based biases in general, and how AI may contribute to these in particular. Surveys were conducted with 60 professionals working in different areas of AI and related issues. The findings revealed relatively informed understandings of the concepts of gender, race and ethnicity, whereas a lack of awareness among participants about prejudice. The study discusses that while considering what new technologies bring to society, it is crucial to understand how these new technologies may perpetuate existing social problems. To prevent such developments, it is crucial that those who play a role in the development of these technologies have an informed and ethical perspective towards the reproduction of social inequalities, for building more inclusive societies.