The self-driving car still has a long way to go, I think. And even assuming their numbers of at-fault crashes reduces to almost nil, there still leaves the factor of other non-driverless cars that will still be around and that you still have to factor in to. The introduction of driverless cars would be as slow, if not slower, than the introduction of electric cars. It's been close to two decades and the combustion engine still outpaces the electric motor by at least 30:1, and that's probably a generous ratio. This means that it could take 2 to 4 decades to get to the point where virtually all the vehicles on the road are driverless.
So lets then suppose that Google, Tesla, Apple, or whomever works out the kinks enough to launch. Okay, so
your car might not crash in to another vehicle, but what about other motorists crashing in to you or pedestrians that make stupid decisions? I doubt a driverless vehicle has the capacity to make extreme evasive maneuvers to avoid something like a deer darting out in the road at night as effectively as a human brain can process and react at this point in time.
It seems that just stopping the vehicle in many instances is simply not good enough because of inertia. Sometimes stomping on the brake while simultaneously swerving might be in order to narrowly avoid a collision, and from everything I've seen of the technology thus far, Google doesn't have an answer for it... yet.
The point is, how comfortable will people really be with just letting the car drive when at least 50% of driving is constantly scanning for threats and responding to those threats?
"Reason obeys itself; and ignorance submits to whatever is dictated to it" -- Thomas Paine