If advanced AI could predict and prevent all crime before it happens, should we implement it knowing it requires constant surveillance of everyone?