Interpretability and explainability from an ethical AI perspective