The impact of self-driving cars is both easy and difficult to envisage.
The easy part to understand is how much more efficiently and effectively they'll do the job humans do today. That is, move people and cargo from A to B. However, instead of a driver there'll be space for an extra passenger.
The next, slightly less easy prediction is the shift in ownership of vehicles. What's the point of owning a car if the convenience of one always being available to take you somewhere can be met by a service? And for a lower cost. The most significant question about this aspect is whether car-as-a-service will be facilitated by a large number of small companies managing localised fleets of vehicles, or whether a small number of large companies will attempt to own the vehicles themselves and service the world (e.g. Uber, Didi, etc.). My thoughts are the former as it is most likely the fastest way to reach scale.
The most difficult impact to predict, however, are the second order effects that Kevin suggests in his slide deck.
The simple starting point to think about this might be: If the cost of operating a vehicle falls dramatically*; there will be a lot more vehicles operating.
Simply, we will be able to 'do more transporting of things'.
Next, absent a human driver, vehicles can be put to greater use to do things which would have been inefficient to do when a person had to be hired to sit behind a wheel. Therefore, a vehicle can start to do a lot more 'things'. Their capabilities will be free to expand. Then two more important questions can be asked:
1) What 'things' can we now move that we couldn't really do efficiently before?
Companies like Dispatch.ai perhaps provide some clues. They're building vehicles to take care of small, last mile supply chain deliveries. Perhaps they'll be competing with drones in the long run, however, at scale, could they do a better and cheaper job than a human delivery person? There are still probably many more vehicle types, like this, out there that are yet to be imagined.
2) What can we now do because of all this cheap transportation?
Suggestions like 'meeting cars' and 'sleeper cars' which Kevin makes in the deck seem very much a likely possibility. Whilst simple sounding concepts, what would impact be? The levelling out of housing prices because commuting became less of an issue? An increase in people living 'nomadic' lifestyles; both working and living anywhere they wanted because it was cheap to do so?
What this points towards, in my opinion, is a future where we will increase our ability to move 'atoms'. Physical stuff. Much like our ability to move data and information digitally increased exponentially over the past few decades due to computing and the internet.
It might take a while. I'm not sure if there's an equivalent Moore's Law in the world of autonomous vehicles. Absent that then it's difficult to see just how quickly this future might arrive; inevitable though it is.
*For a fall in costs to be significant enough, it would likely have to involve a combination of efficiency gains provided by autonomy AND a drop in operating costs achieved through renewable energy like Solar.