Quote:
Originally Posted by Wormius
There is still a load of Trolley Problem ethics problems to work through. Do you let the AI allow a car mow down an 80 yr old instead of a 35 year old, or does the car hit the old Buick instead of a Lamborghini cause of insurance costs? There is something to be said about the randomness of split second human decisions. They are less prone to manipulation.
|
I don't think you give the car the info to make those decisions. What if the 80-year-old is Tina Turner and the 35-year-old is Ben Shapiro? What if the old Buick is driven by an idiosyncratic billionaire whose family sues for the cost of a fleet of Lambos? Those judgment calls are best not made at all and you should just program the car to do the least damage possible.
Ideally, self-driving cars will be networked to avoid almost all collisions except in the case of serious mechanical failure. Even then you could get other cars to help mitigate the damage. Take one for the team and get paid out by the other car's owner or insurance. They'll be happy to do it if it saves them overall.