Institute for Ethics in AI Lunchtime Research Seminar (Wednesday - Week 7, HT26)

Abstract: While much attention has been given to questions of algorithmic fairness in terms of bias, relatively little attention has been given to broader aspects of fairness including the use of data from one context to reach decisions in another. The increasing ability of big data to reveal connections between apparently unrelated data points has resulted in various regulatory and legislative initiatives to control the use of such connections. This paper argues that these initiatives are so far too vague to provide guardrails in specific instances, but also that they fall wide of the mark. Instead, I propose a new framework to guide generalised or predictive decision-making in its application to individuals.

Register