Explainable artificial intelligence (XAI) post-hoc explainability methods: risks and limitations in non-discrimination law

Organizations are increasingly employing complex black-box machine learning models in high-stakes decision-making. A popular approach to addressing the problem of opacity is use post-hoc explainability methods. These methods approximate logic underlying with aim explaining their internal workings, so that human examiners can understand them. In turn, it has been alluded insights from be used help regulate learning. This article examines validity these claims. By examining whether derived post-model deployment prima facie meet legal definitions European (read Union) non-discrimination law, we argue explanation cannot guarantee they generate. Ultimately, explanatory useful many cases, but have limitations prohibit reliance as sole mechanism fairness model outcomes way an ancillary function, inadequacy Non-Discrimination Law for algorithmic decision-making demonstrated too.