Abstract:
Across the world, artificial intelligence (“AI”) is increasingly used to automate decision making for access to pivotal socioeconomic opportunities such as university applications and credit ratings. Given recent high-profile examples of algorithmic bias in those areas, it is timely to consider what it means for AI to be fair and how to achieve that. This article spotlights the growing field of algorithmic fairness, which seeks to directly incorporate fairness into AI algorithms, and identifies two key findings that have significant implications on AI governance. The authors also examine whether existing legal and regulatory frameworks in Singapore, the EU, the US and China can adequately regulate AI decision making, especially in light of the two important findings. The article concludes with several recommendations for how the algorithmic fairness field can further contribute to the ongoing development of AI governance.