‘Ai Theory of Justice’: Using Rawlsian Approaches to Legislate Better on Machine Learning in Government

Abstract

Policymaking is increasingly being informed by ‘big data’ technologies of analytics, machine learning and artificial intelligence (AI). John Rawls used particular principles of reasoning in his 1971 book, A Theory of Justice, which might help explore known problems of data bias, unfairness, accountability and privacy, in relation to applications of machine learning and AI in government. This paper will investigate how the current assortment of UK governmental policy and regulatory developments around AI in the public sector could be said to meet, or not meet, these Rawlsian principles, and what we might do better by incorporating them when we respond legislatively to this ongoing challenge. This paper uses a case study of data analytics and machine-learning regulation as the central means of this exploration of Rawlsian thinking in relation to the redevelopment of algorithmic governance.