I disagree, though I must note that autonomous cars aren’t the main point of this piece. Autonomous cars that communicate with one another would be able to drive in much more compact formations. They don’t need stoplights or stop signs. Basically, they can make much more efficient usage of existing roads. So even if traffic doubled or tripled on them, they wouldn’t make congestion worse. (One example reference is https://www.cs.utexas.edu/~aim/.)

On the other hand, they will remove the need for nearly all parking in the central area of cities as most traffic becomes transportation as a service. This allows you to do better things with all those parking lots.

The move to autonomous transportation as a service has another effect as well, people can get the vehicle that fits the need of their trip, not the one that fits some need they have once each month or less. We all know that most of the cars on the road have a single driver in them and they are all far too big fur that purpose. Most of the cars around you have 4+ empty seats and a lot of unused storage. If you do autonomous transportation as a service, the people commuting to and from work can call on little pod vehicles that seat a single person and have just enough storage for the laptop bag they carry with them. I expect that 80–90% of the active fleet will be those small vehicles. They take up less room and create less congestion. They could also drive in tight formations so you could probably have 6+ in the space taken up by the SUVs and pickups that dominate the roads today.

I also think that there are very cool possibilities for specialized “sleeper” vehicles as well. They probably wouldn’t be any bigger than many of the cars on the road today, but they could be like a little hotel room on wheels that you use when you want to go further. Anything within a 12-hour drive doesn’t require a flight anymore, just have the transportation happen while you are sleeping.

I understand the argument that when cars are autonomous that more people will take rides. That leads to the belief that roads will get more congested. That’s only the first-order effect though and I think it matters less than all the higher-order effects that pop out when people really start to think of the impact more deeply.

Back to the main point of the article though, it isn’t at all clear how long it will be before we see these on the roads in large part because the computing power we need to develop and run them just isn’t happening because the exponential growth in computing speed has basically ended.