Costs would at least be lower, which at least opens the possibility of it becoming cheaper. Sure, it's possible to conceive of a dystopic scenario where GooUber has such a monopoly that they can just keep prices arbitrarily high and disconnected from costs, but I think that's pretty unlikely.

It seems hard to say for sure that costs will be lower. The cost of a fully-autonomous/self-driving car is still unknown. Even if google is currently at the 90%, it's well-known that the last 10% is always the hardest. That last bit is especially important when you consider the consequences of imperfection in this venue.

He might be implying that the cost to rent a car, for 30 minutes, twice a day, would be cheaper than owning an, idle, car 24 hours a day.

Or, sharing the ride with three other commuters could bring the total cost of ownership to ~15 minutes per day.

I meant that the incremental cost of paying a human driver would be gone. The fixed cost of inventing and developing the self-driving technology would be minimal because it is amortized across all self-driving cars.

Fewer accidents drives down the insurance costs.

but the cost of the vehicle increases, which would have an unhappy effect on premiums.

All of that extra cost was R&D, which is currently being funded by Google's advertising. While I have no idea what Google's business model will eventually be with self-driving cars, at the end of the day it's just software and some cameras that can be mass-installed on most any car.

While the cost will initially be higher, it likely won't be much, and will eventually be negligible.

I'm not sure the sensors are negligable...

lower insurance. no salary.