Will AI replace truck drivers? What the record shows (2026)

If you drive a truck for a living, or run a fleet, you have heard the forecast for a decade. Software is coming for the cab, and the long-haul driver is first in line.

The obvious reading is that the clock is running out. The record reads differently. Driverless trucks now haul paid freight, and the official outlook still has driver jobs growing.

Start with that outlook. The Bureau of Labor Statistics projects heavy and tractor-trailer truck driver jobs to rise from 2,221,200 in 2025 to 2,305,000 in 2035, a 4% increase [source].

What “will AI replace truck drivers” actually asks

Will AI replace truck drivers asks whether software can take over the job a carrier pays a driver for: moving the load, and answering for what happens on the road.

Steering on a highway is the part software already does on some lanes. The harder part is who carries the blame when something goes wrong. That is what this post tests.

Will AI replace truck drivers by 2035?

Not on the official forecast. BLS projects a gain of 83,800 heavy truck driver jobs over the decade, and about 214,500 openings a year [source].

The agency’s job outlook names economic growth and online shopping as the drivers of demand [source]. It does not mention self-driving trucks at all.

These are forecasts, not measurements, and TIN did not check them. Compare customer service representatives, where BLS expects a 5 percent decline as tasks “continue to be automated” [source]. For truck drivers, it sees no such effect in this decade.

When will AI replace truck drivers?

For one lane, driverless trucks are already running. Aurora launched what it called America’s first driverless class 8 commercial trucking service between Dallas and Houston in the week of 1 May 2025, with Uber Freight and Hirschbach Motor Lines as launch customers [source].

By September 2026 the company said its system had passed 500,000 commercial driverless miles [source]. It says it remains on track to exit 2027 with more than 1,000 driverless trucks in operation [source].

These are the company’s own figures in an SEC filing. TIN has not checked them. Set them beside the 2,221,200 drivers BLS counts [source] and the scale of the question becomes clear: a target, not a takeover.

The revenue says the same. Aurora reported third quarter 2026 revenue of $3 million across driverless and operator-supervised loads [source].

Will AI replace CDL drivers?

Not yet, even on that lane. Less than three weeks after launch, Aurora moved its human observers from the back seat into the driver’s seat [source].

Its CEO said the truck maker Paccar “requested we have a person in the driver’s seat, because of certain prototype parts in their base vehicle platform” [source]. He said the change was “not required to operate the truck safely” and that “this observer will not operate the vehicle” ([source]).

Read that as an operator, not as a skeptic. The software was driving. The partner that builds the truck still wanted a person in the seat. The seat is a liability question before it is a technology one.

Who is to blame when the automation is driving?

On the records TIN holds, the human in the seat. The NTSB found Uber’s test car detected Elaine Herzberg 5.6 seconds before impact, and that its design “precluded activation of emergency braking,” relying on the operator instead [source].

The probable cause it named was the operator’s failure to monitor the road while distracted by her phone [source]. She pleaded guilty to endangerment and was sentenced to three years of supervised probation ([source]). Uber, as a corporation, was not charged.

The courts have started to move some of that weight to the maker. A Miami federal jury found Tesla 33% responsible for a fatal Autopilot crash, and the final judgment added $200,000,000.00 in punitive damages [source]. Tesla has said it will appeal.

The regulator acted on the same fault line. Tesla’s recall 23V-838 covered 2,031,220 vehicles after a finding that Autosteer’s controls “may not be sufficient to prevent driver misuse” [source]. Tesla filed it while not agreeing with the agency’s analysis.

The proof

None of TIN’s case files is about a truck, and this post claims no trucking outcome. What they hold is the liability record for automated driving, which is the question a driverless truck has to answer.

TIN’s case file on Uber quotes the NTSB’s probable cause and the guilty plea line by line: the first pedestrian killed by a self-driving car.

TIN’s case file on the Benavides verdict rests on the court’s own final judgment and flags the two figures that come from press: a Miami jury found Tesla’s Autopilot defective.

TIN’s case file on the Autopilot recall quotes Tesla’s own filing and the regulator’s closing record: 2,031,220 vehicles fixed over the air.

The wider pattern of AI job claims is argued in will AI replace call center agents. None of these files counts a truck driver job lost to AI.

What each record shows

The bottom line

AI will replace the driver’s seat before it replaces the truck driver, and the gap between the two is liability, not software. Aurora’s trucks can drive the I-45 lane, yet its own truck maker wanted a person in the seat, and the official US outlook still adds drivers through 2035. On every automated-driving record TIN holds, blame landed on whoever sat behind the wheel until a court reached the maker.

The transferable test for any automation claim: ask who answers when it fails, and you will know whether the job is gone.