I think you’re falling into a bit of a trap here: perfect is the enemy of good. Not everything has to be automated, instead of growing pains, there can also be gains.
Remember, we are currently aiming to get these vehicles on the road, alongside regular drivers. They use sensors and computer vision to read street signs, detect people etc., all with the reaction speed of a machine. What if the in-between product is simply a better driver with faster reaction times? That is the current goal, really - no one wants to automate everything, simply because that wouldn’t be feasible anytime soon.
Yes, again, we’re not there yet and these things are far from perfect. But let’s first just aim to get them good enough, and then maybe just a little better than your average driver.
As for the your proposed business model: we have capable drivers now, why do these business models don’t exist right now? Why is there no fast lane that allows me pay to get to my destination faster? What would the technology of driverless cars introduce that would enable these regulations?