AI Assurance
The problem
A successful demonstration does not establish that an AI system is appropriate for its intended use. Important questions can remain about affected people, information, reliability, controls, ownership and the consequences of failure.

