AI Glossary Term
What does “Explainability” mean?
The degree to which a model's behavior can be interpreted and explained to humans.
How Explainability works
Example
A credit model denies an application. An explanation tool shows that high debt-to-income ratio and short credit history most influenced the decision. The applicant and a reviewer can check those facts and correct any errors.
Why it matters
People need reasons to trust, audit, or challenge automated decisions. Explanations also help developers debug models.
Common questions
Are explanations always accurate?
Not always. Some methods approximate the model, so an explanation can be incomplete or misleading.
Is explainability the same as interpretability?
Often used interchangeably. Interpretability usually means the model itself is understandable, while explainability may rely on added tools.
Do laws require explanations?
In some contexts, yes. For example, certain credit and data-protection rules give people a right to information about automated decisions.