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.