It’s funny that the trolley problem thought experiment which had become so déclassé to talk about — man, you were just so uncool — is highly relevant to problems faced by driverless cars.
Thought experiments are frowned upon because they often force one to think about unpleasant eventualities in the service of making better choices, but they are really useful for discovering the problem space and the consequences of various solutions.
Of course one has to be very careful to avoid false dichotomies and other framing errors which can creep in.
I read the article and I don’t agree with the premise. The problem is relevant if you assume human levels of failure with driverless cards. This will indeed be true in the near future as as algorithms get better and the percentage of driverless cars on the roads reaches a certain threshold, the error rate would be so low that it would render the whole question moot.