The problem isn't getting people into the same car. The bigger problem is getting people into the same car at the same time. Sure, self-driving cars can work with each other in ways human drivers can't to minimize congestion, but until you start removing cars from the road, it will still be an issue.
Or perhaps Google Maps will do that. It supports Uber internally as a travel option, no? Wouldn't it be interesting if that dropped away and was replaced with Google's self-driving car scheme one night?
Quite frankly, at those prices, I'm not sure either has an appeal greater than my own car, sitting in my garage, available whenever I want.
I'm extremely sceptical that all the people I see driving on the freeway every day would rather sit in a Prius with three strangers twice a day on their morning and evening commutes.
I think driverless cars are the future of transportation, I'm just not certain that driverless cars and car sharing will be married at the hip from day one. There is lots to figure out to make car sharing appealing at a larger scale than it already is. Most of the things I see suggested just don't have enough appeal to convince me that people will change their current behaviors for them.
That's a bit easier to solve. Congestion pricing along with economic incentives to employers who employee remote workers. Traffic is an externality; if you absolutely need workers on site somewhere in a dense urban area, you can do so, but you're going to pay for the privilege of all that infrastructure you're tying up.