Help the user understand how often the AI system may make mistakes.
Limit disruptive changes when updating and adapting the AI system’s behaviors.
Immediately update or convey how user actions will impact future behaviors of the AI system.