Walker’s mistake is in the following assumption:
“more and smaller autonomous vehicles would increase the overall amount of vehicular travel.”
The problem is that this is linear thinking and (seemingly) assumes there does not exist a demand within the current stock of car commuters to use a bus.
However we know this is not true. When the cost/benefit of commuting by car vs. commuting by bus changes, ridership changes. If we have more, smaller, more frequent buses/shuttles (which lowers the cost, both actual and perceived, of commuting by bus), many people who are currently choosing to commute via single passenger vehicles will choose buses/shuttles.
The net effect will be a reduction in the overall amount of vehicular traffic.
I’ll leave the math to the reader, but roughly 75% of people commute to work (in the USA) as a single passenger vehicle.