In May 2023, Scotland launched CAVForth, billed as the UK's first full-size autonomous bus service, five self-driving vehicles running a park-and-ride route into Edinburgh, backed by £4.35 million in government funding. In February 2025, after 21 months, it shut down. The stated reason was low passenger numbers. It had run the whole time with two staff aboard each bus: a safety driver behind the wheel and a second attendant in the cabin.

Bus drivers are rated High risk, with a horizon toward 2038. That classification is defensible, but the timeline matters: 2038, not 2028. The technology to remove a driver from a city bus does not work reliably yet, the pilots that have tried keep either closing or keeping a human behind the wheel, and the industry is simultaneously in the middle of a driver shortage severe enough that agencies are cutting service they can't staff.

Europe's bus and coach operators reported roughly 105,000 unfilled driver positions in a 2024 industry survey, up 54% in two years. In the United States, an estimated 43% of transit bus operators are over 55, and school bus employment was running nearly 10% below its 2019 level as of late 2025. The pressure to automate this job is intense. The capacity to actually do it is not there.

Key Points

  • Bus drivers are rated High risk with a Low-term horizon toward 2038, a timeline that reflects real automation pressure paired with a technology that is not yet capable of removing the driver from an urban route.
  • Scotland's CAVForth, the UK's flagship autonomous bus project, shut down in February 2025 after 21 months, having run the entire time with two staff aboard each vehicle.
  • Waymo now provides around 500,000 paid driverless rides a week across roughly ten US cities, but full-size buses, carrying dozens of passengers, some standing, some needing assistance, are a materially harder autonomy problem than a robotaxi.
  • The driver shortage is acute: roughly 105,000 unfilled bus and coach positions across Europe, 43% of US transit operators over 55, and 71% of US transit agencies reporting they have delayed or cut service due to staffing.
  • The U.S. Bureau of Labor Statistics projects just 1% employment growth for bus drivers through 2034, a stagnant profession rather than a collapsing one, with a median wage around $48,370.

What a Bus Driver Actually Does

Operating the vehicle is the visible part. The rest of the job is passenger management: helping riders with mobility devices board and secure, de-escalating conflicts, handling medical emergencies, enforcing fare and conduct rules, and making judgment calls when a route is blocked, a road is flooded, or a passenger is in crisis. A bus is a moving room full of people who are the driver's responsibility, and much of what makes the job hard to automate is the room, not the driving.

The Pilots That Keep Closing

CAVForth is the clearest cautionary case, but the pattern is broad. Beep's NAVI service in Jacksonville, launched in 2025 as the first fully autonomous public transit system in the US, runs 14 small electric vehicles on a 3.5-mile route, weekdays only. Karsan's autonomous e-ATAK, the relative success story, operates across roughly a dozen European locations at Level 4 autonomy, meaning self-driving within tightly defined conditions, still typically with a safety operator present during rollout. None of these is a full-size city bus running a normal urban route with no one aboard responsible for the vehicle.

THE ROBOTAXI GAP

Waymo's robotaxis are genuinely driverless, at scale: roughly half a million paid rides a week, over 220 million fully autonomous miles, and a crash record its data puts well below human drivers. Buses are years behind that, and the gap is structural. A robotaxi carries one or two passengers who buckle themselves in. A bus carries dozens, some standing, and cannot execute a hard emergency stop without injuring them. It has to kneel for wheelchairs, wait for slow boarders, and handle a medical emergency in row 12. The autonomy problem scales with the number of people whose safety the vehicle is responsible for.

The Shortage That Won't Wait

The demographic and working-conditions pressure on this workforce is severe and immediate. A US transit advocacy analysis found 43% of bus operators over 55, with a wave of retirements approaching and rising rates of assault and harassment on the job driving people out early. An APTA survey found 71% of transit agencies delaying or cutting service because they could not hire enough operators, and 92% reporting hiring difficulty. Europe's operator shortage is projected to worsen through 2028. This is a profession that cannot currently find enough people, which is a very different situation from one where AI is pushing people out.

What AI Is Actually Doing Now

The real, deployed AI in transit today is route and scheduling optimization. Platforms like Spare and Via run dynamic microtransit and paratransit systems that group trips, adapt routes in real time, and optimize driver shifts, with agencies reporting paratransit cost reductions around 30%. This reduces the number of driver-hours a fleet needs to deliver a given level of service, and over time that does mean fewer shifts. But it is driver optimization, not driver removal: every one of those trips still has a person behind the wheel, and the technology's effect on headcount is gradual, not a step change.

The Labor Market Reality

The U.S. Bureau of Labor Statistics projects 1% employment growth for bus drivers from 2024 to 2034, slower than average, against a base of about 546,100 jobs and a median wage near $48,370. That is the profile of a stagnant occupation, not a disappearing one. The near-term risk to an individual bus driver is far more about service cuts, agency budget pressure, and the hard working conditions driving turnover than about an autonomous vehicle taking the route.

How to Use AI as a Bus Driver or Transit Operator Now

For drivers: route-optimization and predictive-maintenance systems are tools that make shifts more predictable and vehicles more reliable. Driver-assist safety features, collision warning, lane keeping, automatic braking, are arriving on conventional buses first and are worth engaging with rather than resisting.

For agencies: the deployable win right now is scheduling and demand- responsive routing, which stretches a short-staffed workforce further. Full autonomy is a decade-plus procurement question, not a current staffing plan.

What I Think

The 2038 horizon is, in my read, roughly right, and the High-risk label is doing a lot of work that the timeline then walks back. The direction of travel is real: robotaxis have proven driverless operation is possible, and the money and attention will keep pushing toward larger vehicles. But the pilots keep demonstrating how far there is to go, and they keep failing on economics and ridership before they fail on technology.

What I'd actually watch for this profession over the next decade is not the autonomous bus. It's whether transit agencies, squeezed between a shortage they can't fix and budgets they can't grow, use partial automation and demand-responsive routing to run permanently leaner driver rosters. That's a slower, quieter erosion than a self-driving fleet, and it's the one already happening.

"The autonomous bus that was supposed to prove the future closed for lack of passengers, with two employees still aboard. The shortage it was meant to solve got worse the same year."