Algorithmic fairness challenges linked to feedback loops over time
Algorithmic Fairness Over Time: Advances & Prospects
Machine Learning
Summary
Decisions made by computer algorithms can influence the data they receive in the future, which sometimes causes unfair outcomes that get worse as time passes. The authors explain that ignoring these effects can introduce bias and make fairness problems harder to detect or fix. They classify different ways fairness can be considered both at a single moment and over time. The paper also points out open questions and challenges to help researchers improve how fairness is understood when data changes in response to past decisions.
What this means in practice
- •For policy makers: Design regulations that account for changes in fairness due to feedback effects in deployed algorithms.
- •For data ethics teams: Monitor and assess fairness of models continuously to identify bias introduced by evolving data caused by prior decisions.
A survey. It maps existing work.
Authors
Miriam Rateike
Abstract
In the presence of feedback loops, algorithmic decisions at one time step can influence observed data in subsequent steps, potentially introducing selection bias or shifts in data distribution. Research has demonstrated that ignoring these effects can lead to unfairness, which may worsen over time as the model remains deployed within society. This raises important questions of how to measure and ensure fairness of algorithmic decision-making models over time. In this short paper, we present a taxonomy that distinguishes between different types of static and temporal fairness notions in recent advancements. We then suggest prospects for future research by identifying common implicit temporal assumptions, open questions to enhance our temporal understanding of fairness, and other open challenges.