There's a simple way around the _ethical_ part of this dilemma: give the self-driving car the _ability_ to take different ethical approaches (e.g. preserve occupants over onlookers) but then require the operator/owner to select the ethical setting.

This means the developers just enabled the functionality, and it was the operator that chose to be selfish.

That said, none of that deals with the difficulty of making time-critical decisions. For that, you should probably read my thesis...