I got so annoyed when family members would say "just rent a car and come down to visit us". It's not cheap, it's not easy, and it's not fun.
Renting a car in Stamford, CT is cheap, ~$70 a day.
Zipcar now have differential pricing (i.e. more expensive on weekends), so a 24 hour rent Mon - Fri is under $100.
Expensive, you may say. But... I pay only $70 per year for a full insurance waiver, all fuel is included, and there are no maintenance costs or other hidden costs.
Last Zipcar I got I used almost a full fuel-tank worth in a 24 hour period. Subtracting fuel from the price put the days rental at about $40. And there's a pod across the street from my apartment (and about a zillion other pods nearby, here in Berkeley CA).
I have friends with a 2-kid family in Manhattan and no car. They use Alamo, and claim it's cheap and easy.
Renting is expensive. Owning a car is also expensive, with more 'hidden' costs that people frequently ignore.
When people have already bought a car and paid a flat insurance rate, the marginal cost of another car trip is very low. Thus, people who own cars don't really make a choice very frequently. Their choice has already been pretty much made once at the outset.
Because carsharing services will almost undoubtedly be the first deployments, self-driving cars cannot come quickly enough for me. I think the net effect is that they will drive (har har) more people to adopt carsharing services. Self-driving cars will solve one of the inconveniences of something like Car2go, namely the need to walk some indeterminate distance to a car.
If they can get more people into an a la carte transportation usage model, I anticipate public transit and non-motorized transportation options will benefit greatly. Since trip costs will only be marginal, the choice to do something other than drive will come up routinely.
As self-driving carsharing fleets replace the >90% idle private vehicles of today, we'll need less land for parking. My hope is that on-street land can be given over to wider sidewalks, better quality bicycling infrastructure, and dedicated transit right-of-way, which will make those modes even more competitive.
In part because I've been personally touched by the mass slaughter on our roadways, I left a career in information security to retrain / work as an urban planner focussing on transportation. Oddly, the nearishness of self-driving vehicles is pushing me back in the direction of my former career. I'm bullish enough on the potential of self-driving vehicles to remake the economics and safety of transportation that I'm now pursuing a PhD researching privacy / security aspects of the "smart city"--including on-demand mobility services. If we can get all this suitably right, I think it will be transformative for society.