Dernière mise à jour :
While both aim to make AI decisions understandable, explainability focuses on the reasoning behind decisions, whereas interpretability relates to the transparency of the model's internal mechanics.
Interpretability: clarity about how internal model components (weights, features) map to outcomes - common in simple models (linear regression). Explainability: post hoc generation of human-friendly justifications (why a decision was made) for any model, even black boxes. Governance requires choosing the right balance: interpretable models where possible, and explainability tools where not.
Exemple concret :
A bank chooses a logistic-regression model for credit scoring because of its interpretability (coefficients directly show feature impact). For its image-based fraud detector (a neural net), it uses explainability (saliency maps) because the model itself isn’t inherently interpretable.
