The EU General Data Protection Regulation (GPDR) grants individuals the right-to-explanation, via requiring institutions to provide explanations to individuals that are subject to their (semi-)automated decision making systems. (Quote from [@karimiModelAgnosticCounterfactualExplanations])
This means that in order for a system that utilizes some automated decision making system (such as a Multi Layer Perceptron), need a method to explain the decision that the model made. The problem is that finding explanations for complex deep learning based model is still an active research topic, and it’s not trivial. The topic is called “AI Explainability”.
tags: ai-explainability see also: Counterfactual Explanations in Recommender Systems