Driverless cars are where governments have to take a holistic approach to road safety, rather than a manufacturer-centric one.
All manually controlled vehicles will need automatically invoked mechanisms to prevent being used while the driver is too impaired or attempting risky manoeuvres. And what is to be done about muscle cars, as an all-roads safe-driving regime will be a real killjoy for their owners? Of course, all the law enforcement agencies will want to be able to sideline cars as required, such as for emergency vehicles, which should be considered as movable infrastructure, rather than just another competitor for road-space.
However, another consideration is to look at how large numbers of automated vehicles could benefit from a rethinking of main roads and their feeds. If all cars on such roads are speed maintained and responsive to infrastructure feedback, road planners can make far more reliable decisions about utilisation and traffic management. There is a potential to make huge optimisations to reduce vehicle energy usage and travel times, even having dynamic traffic rerouting.
Publically provided small autonomous vehicles even have the potential to eliminate buses and trains, while providing much more granular adaption to numbers travelling, thus saving substantial energy. If people buy their own compatible vehicles, they can just merge into the traffic streams. Main routes could even provide energy to electric vehicles, thereby reducing battery wear, saving substantial impacts on the environment.
Autonomous, but cooperative vehicles have the potential to overhaul our public and private transport to be fully door-to-door without having to have them tied up in parking for most of the day.
Just as we have luggage that is designed for aircraft lockers, we could have personal luggage that contains our own portable office that we wheel up to, snap into, and open up in, a transport vehicle to power it, and even wheel into an office pod shell, at work or home, that does the same. Bring your own office everywhere!
Of course, it will take some time for insurance companies to gather the actuarial data to determine their risk, but if 'driving' fully autonomous vehicles, it won't matter what age you are, but only how far you travel and where.
Given the amount of computer power and data already being thrown at providing a vehicle's autonomous functionality, I would expect that insurance companies will want your telemetry data to calculate how risky you are. Besides the speedo, tachometer and instantaneous power-usage/fuel-consumption meter, if we still need to see them, there can be an instantaneous insurance risk meter.