Once the AI is working, surely the corporation can just ask it to come up with an inexpensive accountability sink for trolley problems? Perhaps it will suggest hiring an impeccably credntialed AI ethicist to legally opine that it has the most ethical trolley problem solutions?