M. Vladimirova

Position: Fairness Failure in Generative Models is an Evaluation Problem

This position paper argues that fairness failures in generative models, albeit driven by multiple factors, are ultimately stemming from an evaluation problem: fairness findings are rarely comparable across papers or actionable for deployment decisions.