Algorithmic Discrimination and the Law: Regulating Bias in AI Decision-Making

Abstract

Algorithmic discrimination is a recent phenomenon which has begun to drastically influence outcomes in the judicial system, healthcare, social services as well as employment. These algorithmic systems are mostly trained on historical data which is biased or based on irrational decision models, which can therefore worsen the existing social gap. This paper attempts to analyze the existing issues concerning algorithmic bias and assess how the current anti-discrimination laws could be revisited for incorporating provisions to regulate the algorithmic decision-making process. It determines whether the prevalent legal frameworks are sufficient to tackle the hidden nature of discrimination led by AI. This study adopts a comparative legal analysis combined with a case study approach, examining legal frameworks and judicial developments across different jurisdictions to evaluate regulatory responses to algorithmic discrimination. Reliance will be placed on relevant case laws and policy developments across multiple jurisdictions in order to identify enforcement gaps and opportunities for regulatory innovations. Furthermore, the paper attempts to explore the potential of modern AI driven tools for instance, impact assessments, algorithmic transparency mandates as well as audit requirements. Lastly, the paper suggests a precautionary and rights-oriented approach towards governing AI which is based on the principles of human dignity, justice, and accountability. Without sufficient legal oversight, algorithmic systems can replicate the discrimination and biases which were intended to be eliminated.