Robustness and explainability of Artificial Intelligence