Self driving cars will help congestion, but probably not as much as people think they will.
First, we need to think about the sources of congestion. There are a few, and only some are human caused.
The human caused sources include traffic accidents and unsound driving practices. Obviously accidents can mess things up on the road very significantly, but the effects are transient and infrequent. Poor driving probably leads to some congestion also, such as when people fail to merge despite warnings.
The other, major source of congestion is simply a problem of physics, specifically the conservation of mass. Traffic can be thought of as an incompressible fluid, so that mass flux must equal zero over any closed loop (in steady state). More simply stated, there's only so much stuff you can pack into a given space. The only way self driving cars can alleviate any of this congestion is by making cars drive closer together, thus compressing the fluid. Even if they halved the safe driving distance, the gains would only be marginal, because each car needs to slow down to accommodate any new cars entering the freeway. Therefore, traffic jams may be more enjoyable due to our ability to read while enduring them, but they certainly will still exist.
I wonder whether it will affect the flow of traffic in cities. That is, will self driving cars stop at red lights if they're certain that no pedestrians, other cars or obstacles are present? This could speed up travel times, and increase fuel economy greatly.