Think about it; You need to pay a driver something like $40k each year; While the car itself may cost $40k to purchase and will last for 5-10 years.

Having human drivers also modifies the way you deploy your taxies. The driver must return home at the end of each day, so intercity trips don't happen. The driver needs to be paid even if the taxi is just sitting waiting for a passenger, so you want to closely match the number of active taxies to the number passengers. This also means electric taxies are unattractive for for manned taxies because your taxi shouldn't be stopped for long enough to charge between rides.

But if you move to self driving electric taxies you get rid of idle costs. It costs almost nothing to have a taxi which doesn't have passengers, so you can overstock the taxies and give time for them to charge between rides. Since you have extra taxies, you can also afford to leave them in remote areas where you know you will have return passengers in the future.

Basically, a self driving taxi doesn't have the same cost model, so it is reasonable to expect a change in the usage patterns.

Different costs can lead to radically different consequences. Just imagine how smart phones or tablets would play out without, say, cheap touch screens. People wouldn't care much for them.

Or imagine if power was much cheaper, then we could run eg desalination for almost free, and no coastal area or island would ever have any problems getting fresh water.

A well-planned system of autonomous cars can anticipate your needs to a high degree of accuracy, either as an individual or as a group.