I eagerly clicked on this analysis of the effect robotaxis will have on traffic, but I have to say I came away from it dissatisfied.
If and when self-driven vehicles become widespread, research and common sense both suggest they will cause car use to spike. The resulting logjams would be calamitous for cities’ quality of life as well as their economic health — unless public leaders enact preemptive policies to keep people moving.
To understand why AVs will induce additional driving, consider the reasons that American robotaxi companies have spurred so much excitement among their investors and early customers. Self-driven vehicles offer superior car trips, particularly when compared with regular ride-hailing services or taxis. Today’s robotaxi passengers can set the temperature and select their favorite music, and they need not worry about a driver eavesdropping on private conversations or judging off-key singalongs. For regular users, robotaxi companies like Waymo, Zoox and Tesla also provide reliable consistency across their fleets of identical vehicles using the same software and hardware.
These advantages can be substantial. Case in point: Waymo’s Bay Area riders pay a premium of at least 10% to take a robotaxi instead of human-driven ride-hail.
How do people respond when an experience improves? They want to do more of it.
It’s a safe bet that AV customers will take more and longer car trips than they would have if they (or another person) sat behind the wheel. Some of those autonomous rides would have otherwise occurred on transit or by bicycle, or perhaps never happened at all. Given the pleasantness of traveling in an AV, workers will also tolerate longer car commutes, catalyzing sprawl (and creating yet another headache for America’s beleaguered transit agencies, which rely on density to provide efficient service). All of these behavior changes point in the same direction: more miles driven.
Then there is the issue of “deadheading,” industry parlance for the driving that occurs without any passenger inside the vehicle, as when an AV is en route to a pickup, waiting for its next summons, or headed toward a recharging station. As of late 2025, deadheading represented almost half of Waymo’s total miles driven in the Bay Area.
In a recent academic paper, University of Texas-Arlington engineering professors Farah Naz and Stephen Mattingly reviewed 26 prior studies evaluating how autonomous vehicles will affect total vehicle miles traveled (VMT). They estimated a VMT rise of 6% attributable to AVs, with the anticipated increase a bit higher for privately owned AVs and slightly lower if robotaxis remain the dominant format.
Each vehicle mile that AVs induce will impose costs on society, including additional crashes (since any car trip carries a degree of risk) as well as pollution from tires and the generation of electricity.
I’ve been thinking about the effect that autonomous vehicles will have on traffic for over a decade now, and I’m definitely open to the idea that the net effect will be worse. I think the logic cited in this article is generally sensible. What I was disappointed in was the lack of data. It was all assertion and basically no numbers. The one number that is cited is the estimated six percent rise in vehicle miles traveled, which honestly doesn’t sound that much. Certainly not “the streets will all be jammed and it will be too late to do anything about it” levels of alarm. I was especially disappointed since the author of this piece was the same as the author about this earlier analysis of how “safe” autonomous vehicles are, which I thought had a lot of depth and nuance. (I’ve been thinking about the safety issue for a long time, too.) We have some more recent evidence in a story that is unfortunately paywalled but was in a recent print edition of the Chron – the link headline is “Tesla’s robotaxi crash count reaches 15 since June in Austin. Waymo hits 60” and the story headline is “Tesla, Waymo report new Austin crashes as robotaxi operations face increasing scrutiny”. I do expect things to improve on that end, but given the long history of overpromising, I’m going to keep my expectations modest.