1. I mean, shit, I write software for a living, I know how the sausage is made. And I know how shitty and brittle and error prone most software is. I won't be putting my life in the hands of your shitty non-tested, non-verified, low-bid code anytime soon!
2. Recognizing that there's a HUGE difference between doing self-driving in Arizona on a nice long, straight, smooth highway, in summer-time, in broad daylight, and doing it on a windy, twisty, narrow, back-country road in the mountains, in the middle of winter, with snow falling, a layer of ice on the road already, in the middle of the night. To be fair, humans don't do particularly well in that setting either, but still...
All of that said though, I lean towards what somebody (was it Eric Drexler?) said about how "if a technological advancement isn't literally impossible under the laws of physics, then it is going to happen. The only question is when."
I have a rule that I'm not going to make predictions about the industry until 2025. Nonetheless, it has been a good year for machine learning and that includes autonomous vehicles. The survivors this year are showing more signs of optimism than they have since about 2017, which was a year of failing to live up to big promises that they are finally starting to live up to now.
Yes in a sense where there is fleet of contractors that can take over vehicle any time if something goes wrong.