KJuris: Kate Vredenburgh 'The Limits of Explainability for Reducing Algorithmic Discrimination' - 14 February 2024

The Yeoh Tiong Lay Centre for Politics, Philosophy and Law is delighted to host Doctor Kate Vredenburgh for the fifth workshop in the 2023/24 KJuris programme.

Title

The Limits of Explainability for Reducing Algorithmic Discrimination

Abstract

Proponents of algorithmic decision-making have argued that the use of algorithms can reduce discrimination, against the baseline of human decision-making. One reason is the greater explainability of the models, or the ability to provide information so that people can understand how inputs influence outputs. This talk examines the relationship between discrimination and explainability. I will argue that people are at least as explainable as AI for the purposes of detecting discrimination, and that explanations of particular algorithmic decisions are of limited use in providing legal proof to combat discrimination. These two claims should lead us to think that algorithmic decision-making is not preferable to human decision-making on the grounds that discrimination can be made more transparent and provable.

Author Bio

Kate Vredenburgh is an Assistant Professor in the Department of Philosophy, Logic and
Scientific Method at the London School of Economics. She works on topics in the philosophy of social science, political philosophy, and the philosophy of technology.