I think the importance of the trolley problem is overstated.
David Chase

The problem is, though, that (minor) x (at scale) == still a major issue to be dealt with from a regulatory perspective before the cars could even get out on the road and “empirically determine” the legal ramifications of actual liability suits.

There are about 5.5 million vehicular accidents per year in the U.S., and even if self-driving cars reduce that figure by an estimated 90% we are still talking 550,000 crashes per year that are going to have implications for the insurance industry, auto and tech industries, litigation, local law enforcement, and local governments — because the press coverage of Google or Uber’s shiny new innovation plowing into a bunch of school kids is not going to “play well” with local communities (nor stay local coverage for long), even if statistically-speaking, those truly dramatic examples of the Trolly Problem crop up “only” a handful of times per year.