I don't agree. The problem of least 'harm' will come up continually. Moralistic decisions must be made by the car. And even if those incidents do not occur, the software will have to be programmed for these types of decisions.