Algorithmic Explainability

The degree to which the decision-making process of an AI system can be understood and explained by humans. In regulated sectors such as finance it is a compliance requirement, and post-hoc explanation tools such as SHAP and LIME are commonly used.