Ride-sharing isn't going to solve these problems. At most, you fit up to five people in a single car. Most will carry fewer people, perhaps only one or two.
If you try to scale up ride sharing vehicles so they carry more people, you now have to build specialized vehicles that make multi-stop trips where you pick up multiple parties on-demand and drop them off at their individual locations. You've now eliminated the advantages of ride-sharing or taxis--they pick you up where you want and take you directly where you want to go--and managed to recreate a smaller bus. Let's call it a small, self-driving minibus that's not a bus system but a ride share system.
Only it's much worse than a bus, because bus routes are optimized in a way that vehicles picking up and dropping off passengers at what amounts to random locations on-demand never can be. Plus, you're adding even more vehicles to the road in addition to all the other self-driving cars that are running around making deliveries, looking for parking or running off to the outskirts of the city to kill some time as they wait for their human passenger to summon them again, or hustling to try to pick up passengers to make some money for their owners.
Heck, you can't even have them drive on access-limited busways or bus lanes because, by their very nature, they can't stick to dedicated routes.
Consider this: the Siemens S700 streetcar carries over two-hundred people in its usual configuration[0] with a length of just under 95 feet, or the equivalent of around ~6 cars that are almost bumper-to-bumper and trams can be built to be autonomous.
Self-driving cars have their benefits, but they're not an alternative to actual public transport solutions no matter how we try to dress them up. In the end, and as others have pointed out elsewhere in recent years, every new transit idea eventually boils down to a train. Or, in this case, an actual bus.
0. https://assets.new.siemens.com/siemens/assets/api/uuid:3ce5a...