If everyone already has cars, it takes a lot to get people out of cars --- if someone already has a car, taking the bus to go somewhere slower than driving is completely out of the question. Conversely, if everyone doesn't have cars, it takes a lot to get people to drive cars due to the upfront cost of car ownership.
I think more transit-oriented city development is definitely the way to go. But nibbling away at the problem is going to take a long time. I think robotaxis are definitely going to play a part in this: by being compatible with existing car-dependent infrastructure and providing a similar speed and flexibility as privately owned cars, existing car owners are more likely to give up personal car ownership. Once people stop owning cars, they would be more amenable to development that gets rid of parking spots, improves walkability, increases density, implements mass transit, and so on.