Should driverless cars kill their own passengers to save a pedestrian?

[Ben Davis]

To the OP, in my current industry (insurance), this is a pertinent and much discussed question. The insurance industry in the UK is lobbying for more & better automation of driving as it makes claim rates more predictable. I think we can discount the idea of the auto-driver turning control over to a human driver in the case of emergencies because in many (most?) incidents, there's simply wouldn't be enough time for the switch-over to give the manual driver thinking and action time. I think the answer lies in speed control, collision recognition and braking tech: if the car's going at a speed which is likely to reduce damage to anyone it hits (e.g. 30mph or below), collision recognition is rapid (possibly predictive) and brakes are improved so to rapidly reduce speed (therefore the force of the collision), many lives would be saved thus removing the need for a car to do anything but take minimal collision avoidance action. It would be impossible to avoid all injury but at least more people would live.