Fatality rates are what's recently been discussed, e.g. related to the Tesla AutoPilot crash. If you switch to measuring "accidents", two big concerns pop up:

Fatality rates are what's recently been discussed, e.g. related to the Tesla AutoPilot crash. If you switch to measuring "accidents", two big concerns pop up: 1) who defines what an accident is? and 2) if we get autonomous cars that end up in less minor accidents but same or more number of fatalities, is it really worth it? We could extend it to "fatalities and accidents where someone is left permanently disabled", which is pretty unambiguous and does measure the most important factors. But I'm afraid (or rather, glad) that this won't give you an orders-of-magnitude increase in statistical samples as you get when counting all accidents.