AI explainability

Explainability means being able to answer "why did the AI decide this?". Some models are easy to explain (a decision tree); others, like deep neural networks, act as a "black box" and need specific techniques to show which factors weighed most.

The AI Act does not always use the word, but it requires it in several ways: Article 13 asks high-risk systems to be transparent enough for users to interpret their outputs; Article 14 requires the person overseeing them to understand them; and Article 86 gives people affected by a decision based on an Annex III system a right to an explanation of the role the AI played.

Documenting how the model works in a model card is the most practical place to start.