Help the user understand what the AI system is capable of doing.
Help the user understand how often the AI system may make mistakes.
Allow the user to globally customize what the AI system monitors and how it behaves.