ABSTRACT
Machine learning (ML) systems, increasingly deployed in high-stakes decision-making, inherently produce uncertain outputs that can lead to unlawful discrimination. This article provides the first legal analysis of how predictive uncertainty in ML systems interacts with UK anti-discrimination law under the Equality Act 2010. Employing a decision-theoretic framework, the article distinguishes between aleatoric uncertainty, stemming from irreducible randomness, and epistemic uncertainty, arising from incomplete knowledge from deliberate model design choices. While identifying and justifying aleatoric uncertainty presents a unique challenge, given its reflection of underlying, irreducible risk external to the model, it may introduce a form of indirect discrimination unique to probabilistic systems. Intentional design decisions introduce epistemic uncertainty, which can directly and indirectly cause discriminatory outcomes. Instead of assuming ML is a ‘black-box’ or evaluating solely by outputs, greater legal importance should be placed on the design choices that are embedded within ML systems. Therefore, the article contends that algorithmic discrimination introduces new challenges that current legal frameworks are ill-equipped to address while also demonstrating how some aspects of unlawful discrimination are clearer in ML settings than previously thought. It also advocates for enhanced interdisciplinary interpretation of anti-discrimination doctrine with support from proactive oversight and regulatory measures beyond individual litigation.
Holli Sargeant, From Estimation to Discrimination: Algorithmic Bias, Predictive Uncertainty, and Anti-Discrimination Law, Modern Law Review. First published: 31 May 2026.
Leave a Reply