A 2026 review of research on automated vehicles estimates that adoption could increase total driving in US cities and metropolitan areas, even as the technology promises greater independence for people who cannot drive.
The analysis in Travel Behaviour and Society combined 195 estimates from 26 published articles. Across shared and privately used automated vehicles, it found an average increase of 5.95% in vehicle miles traveled, the total distance covered by vehicles.
That figure summarizes research estimates. It is not a measured nationwide increase caused by today’s robotaxis, nor a prediction that every city will experience the same change.
It illustrates why the effects of self-driving transport extend beyond replacing a driver. Passengers may gain time and mobility, while cities face more traffic and workers face changes in the jobs available to them.
Driver assistance and driverless service are different
The US National Highway Traffic Safety Administration distinguishes systems that assist a human from those that perform the driving task themselves.
At Level 2, a vehicle can control steering and speed together, but the driver must remain engaged and supervise. At Level 3, the system can drive under specified conditions, with a person available to take over when requested.
Level 4 can operate without a human driver within defined conditions, such as a service area. It does not mean the vehicle can handle every road and weather condition. Level 5 describes automation across all driving conditions.
These categories matter when comparing announcements. As our report on Uber and Wayve’s London rides explains, a service with a safety driver is different from one that carries passengers without a driver in the vehicle.
Access depends on the whole journey
For a person unable to drive, a reliable autonomous service could make employment, shopping and appointments easier to reach. Affordability, coverage and accessible design determine whether that possibility becomes useful transport.
Australia’s 2025 accessibility guidelines for automated public transport address the process of finding and identifying the correct vehicle, as well as boarding and traveling in it.
A car that drives independently may still leave a passenger needing help at the curb. People with different disabilities may need accessible booking tools, clear audio instructions, wheelchair accommodation or assistance that a conventional driver would otherwise provide.
Removing the driver therefore creates service-design responsibilities as well as opportunities.
New jobs may not replace the old ones evenly
A 2023 study of alternative occupations for US heavy-truck drivers examined wages, entry requirements, projected demand and geography.
Many identified alternatives required relatively little additional training or education, but paid less than truck driving. Available opportunities also differed between states. The study explored displacement scenarios; it did not establish how many drivers would lose their jobs or when.
That makes the quality and location of replacement work important. A vacancy requiring a move or a substantial pay cut offers a different outcome from a comparable job nearby.
Automation also leaves operational work behind. Our earlier coverage of robotaxi labor costs examined continuing needs such as fleet support and maintenance. These expenses affect what operators can charge and whether a service can make money.
Convenience can generate more driving
The University of Texas at Arlington researchers behind the 2026 review point to empty vehicle trips, pickup activity and pressure on curb space as issues cities should prepare for.
A vehicle traveling to its next passenger uses road space even when nobody is riding. Easier travel can also encourage additional journeys. Some could replace difficult or impossible trips; others could replace public transport.
The review found a smaller estimated mileage increase for shared vehicles than for non-shared ones. That does not establish a universal outcome for a particular fleet or prove that a given increase in mileage produces an identical increase in congestion.
Safety needs evidence from the service being offered
Automated driving could reduce crashes associated with human mistakes. NHTSA also identifies challenges in demonstrating safety and managing cybersecurity risks.
For passengers and regulators, evidence from the actual operating conditions matters: the roads covered, situations excluded and response when the vehicle cannot proceed. A company’s results in a limited service area do not establish performance everywhere.
Cities can assess expansion against several concrete outcomes: who gains usable transport, how much empty driving occurs, whether safety improves and what happens to affected workers. Counting vehicles alone will miss much of the change.
