Measuring Non-Expert Comprehension of Machine Learning Fairness Metrics

Abstract

Bias in machine learning has manifested injustice in several areas, such as medicine, hiring, and criminal justice. In response, computer scientists have developed myriad definitions of fairness to correct this bias in fielded algorithms. While some definitions are based on established legal and ethical norms, others are largely mathematical. It is unclear whether the general public agrees with these fairness definitions, and perhaps more importantly, whether they understand these definitions. We take initial steps toward bridging this gap between ML researchers and the public, by addressing the question: does a lay audience understand a basic definition of ML fairness? We develop a metric to measure comprehension of three such definitions--demographic parity, equal opportunity, and equalized odds. We evaluate this metric using an online survey, and investigate the relationship between comprehension and sentiment, demographics, and the definition itself.

Cited by in corpus (10)

• Explanation Strategies as an Empirical-Analytical Lens for Socio-Technical Contextualization of Machine Learning Interpretability
• Fairness Through Robustness: Investigating Robustness Disparity in Deep Learning
• PreferenceNet: Encoding Human Preferences in Auction Design with Deep Learning
• ProportionNet: Balancing Fairness and Revenue for Auction Design with Deep Learning
• Technical Challenges for Training Fair Neural Networks
• Can Active Learning Preemptively Mitigate Fairness Issues?
• Counterfactual Explanations for Machine Learning: Challenges Revisited
• Probabilistic Fair Clustering
• Fair Clustering Under a Bounded Cost
• A Sociotechnical View of Algorithmic Fairness