From Fairness Metrics to Key Ethics Indicators (KEIs): A Context-Aware Approach to Algorithmic Ethics in an Unequal Society
31 Pages Posted: 12 Oct 2020
Date Written: July 31, 2020
As machine learning algorithms are increasingly used to inform critical decisions across high-impact domains, there has been rising concern that their predictions may unfairly discriminate based on legally protected attributes, such as race and gender. Scholars have responded by introducing numerous mathematical definitions of fairness to test the algorithm, many of which are in conflict with one another. However, these reductionist representations of fairness bear little resemblance to the real-life contextual considerations, failing to reflect the long-standing debates in ethical philosophy and welfare economics on what it means to be fair. This paper will discuss two gaps: 1) between the assumed simplicity and separability of unacceptable bias and the complexity of the contextual and ethical debates on equality and 2) between the context-driven reality of fairness considerations and the axiomatic, unambiguous formalisation of fairness. We argue that the existing approaches to algorithmic fairness fail to capture the range of important ethical considerations specific to each use case. Thus, we propose a new approach using Key Ethics Indicators (KEIs) to provide a more holistic understanding of whether and to what extent an algorithm is aligned to the decision-maker’s ethical values.
Keywords: algorithmic fairness, algorithmic ethics, fairness, ethical AI, machine learning, key ethics indicators, KEI
Suggested Citation: Suggested Citation