Asking them to whiteboard out software design thoughts, risks, etc, for things we all commonly encounter can be revealing. Things like an ATM, slot machine, bank of elevators, red light cameras, and so forth.
Toy problems get toy answers, you'll learn nothing that way. I couldn't care less about traffic lights and elevators, and I can code circles around many.
I feel like I learned something through your answer.
Elevator control isn't a toy problem which means you haven't thought about it at all.