Enable the user to access an explanation of why the AI system behaved as it did.
Limit disruptive changes when updating and adapting the AI system’s behaviors.
Allow the user to globally customize what the AI system monitors and how it behaves.