Re: country & western singers
> Clearly the list of those at whom an out-of-control autonomous car should aim are as follows, in order of priority
To be honest I find the whole question a bit silly. In my opinion, If the autonomous car is out of control, then by definition, even if it can make a decision on who to kill, it can't actually control the vehicle to take aim.
If the autonomous car still has control over the car, why wouldn't it decide to not kill anyone? It could aim to miss everyone, or just stop. One of the main things autonomous car enthusiasts go on about is how AI self driving cars will be millions of times better than human drivers and their puny animal brains. It will be able to stop faster, it will always drive at a speed fitting the road conditions, predict events better, and be so much safer, etc... i.e. the autonomous car is smart enough to never get into a position to have an accident in the first place.
Assuming all the above is actually achieved in the future (personally as a non techno-fetishist, I am skeptical), the only time the autonomous car has to make a decision about who to kill is if things have gone badly badly wrong. Basically we are talking about a hypothetical situation where the autonomous car still has enough control of the vehicle to use it to kill someone, but for some reason cannot use that control to avoid killing anyone (I don't know, suddenly a large group of people, all in a line and easily distinguishable, instantly appear just metres away from the car? Even then, either the car can hit the brakes and stop, or it can't, and it will skid straight into the group with no control over who exactly it hits).
Barring a malfunction (or it being hacked) that renders the AI car homicidal in intent, I can't actually see a use case for where the autonomous car has to make such ethical decisions.
And on that cheery note. Happy Friday, almost pint o'clock!
Biting the hand that feeds IT
