Lack of transparency or interpretability

The inability of users, auditors or regulators to understand how an AI system reaches its outputs or to inspect its workings.

Scheme

Broader terms