Cet article n'est pas encore disponible dans votre langue ; la version anglaise est affichée.
Automated decisions and the 'right to explanation': making the black box accountable
When an algorithm makes a decision with significant effect on a user, they often have a right to a meaningful explanation and human review.
In credit, hiring, and pricing, where an algorithm alone makes a decision with significant effect on an individual, most data-protection laws grant a right to explanation and human intervention.
A 'meaningful explanation' is not open-sourcing the code but helping the user understand the main factors behind the decision, with a channel to contest and have it reviewed.
At the product level, reserve explanation-text generation, a human-review entry point, and decision logs — turning the requirement into usable interaction rather than after-the-fact patching.