I worked at Google before FIRE, but not on autonomous vehicles (AV), I have no inside info. This is my own opinion.
Like other transformative technology revolutions, autonomous vehicles will change our world in unexpected ways that will seem obvious in hindsight. Few people expected the internet would clobber cable, yet it seems obvious now.
The change will happen much faster than most people expect. The companies working on this are ironing out various issues, but much of the problem is already solved. Once the entire package is working it will deploy rather quickly - just a matter of copying software (zero cost) and installing the hardware (can be done production line). We are hearing about it now but it seems far off, then suddenly there will be driverless car options on the road.
Most people will not own an AV. Large ridesharing companies will maintain fleets of electric AVs, which will be much lower cost to operate than manned ICE vehicles. Competition will ensure these cost savings are passed on to the consumer. The AV manufacturer will assume liability and insurance...I don't see how they can avoid this because their software is effectively the "driver." This will decimate auto insurance companies.
The most dramatic change will be large numbers of families going from 2 cars to 1. The one car will be used for errands, road trips, etc. but will incrementally spend more and more time parked in the driveway. More and more commuters will pool AVs to work - algorithms already exist for batching rides based on time, starting location and ending location. This will greatly reduce costs and allow the AV to use HOV lanes. This will decimate automakers, effectively cutting their market in half or worse -- the single vehicle people own will get less use, hence last much longer.
Congestion will be reduced. Partially because more people will carpool in AVs. Moreso because AVs will be able to communicate with each other in real-time, which means they can broadcast their intentions/actions to surrounding AVs. This will enable AVs to safely link up in virtual "trains" of tightly packed cars - reducing drag and increasing the effective capacity of highways. This will also reduce or eliminate pulsing of congested traffic we see with human drives. This, along with greatly reduced costs (from above) are going to decimate some mass transit systems, especially light rail. Rail in very dense metros will be fine (e.g. NYC) but those in lower density areas (suburbs, sprawling cities) will have a hard time surviving. Bus routes will survive with automated EV busses.
Auto related deaths/injuries will drop rapidly as highway miles driven migrate to AVs. This will decimate personal injury lawyers, but be great for the ~30k people who would otherwise have died in any given year in the US. Life will be much much safer for pedestrians and bikers. Computers don't get fatigued, distracted, enraged, impatient, and 360 degree field of vision (LIDAR and other sensors) means AVs will have no blind spots and be much safer for everyone.
Some security researchers will make a name for themselves finding vulnerabilities in AV software, showing how a hacker could take over or cause a crash. The media will have a field day with the hysteria. These holes will be patched and little will come of it. But expect people to be fixated on the dangers of AVs (because they are new) while ignoring the tens of thousands of people that die because of human driver error. Eventually stats will emerge that driverless cars are much much safer than traditional cars. Anyone remember when people used to be afraid to use their CC online because hackers might steal it? Yeah, no one really worries about this stuff anymore. Same will happen with AVs.