Driverless truck predictions were wrong before. Can this wave finally reduce humans on the road?

In 2016, a hot bet in tech circles was that truck drivers were going extinct. 

Driverless technology seemed unstoppable. Though the number of job losses was in question, the replacement of human drivers looked inevitable. That hasn’t happened. More Americans are truck drivers today. But the question is back on the road.

Aurora Innovation, the publicly-traded icon headquartered in Pittsburgh’s Strip District, announced last month that its trucks have driven more than 100,000 driverless miles on public highways in Texas. 

This week, I got inside the cab and walked the perimeter of one of Aurora’s trucks, on display at Pittsburgh’s Robotics Discovery Day — organized by the Pittsburgh Robotics Network and sponsored in part by the federally-funded New Economy Collaborative. 

Inside, the cab looks like any other — not too different from a truck I rode shotgun in while hitchhiking across the Dakotas 20 years ago. It still has two seats and a steering wheel, which can be operational. But the screens and exterior cameras remind us what this is: a mix of software and hardware that runs complex computer vision and decision-making processes to guide a payload to a delivery point. 

With regulations in mind, driverless cars are becoming more visible in more places. Trucking, which moves three-quarters of all freight in the United States, lags behind, not least of all because a fully loaded tractor-trailer can weigh 20 to 30 times the average car.

Despite this, companies like Aurora are still scaling up.

Alongside its current route between Houston and Dallas, Aurora is adding a second route between Fort Worth and El Paso. The company plans to remove its final “safety observer” by mid-2026, deploying hundreds of driverless trucks in partnership with major carriers like Hirschbach Motor Lines and Volvo.

Aurora’s new announcements raise a fair question. Was the 2016 forecast just truth delayed, or were we wrong about autonomous freight altogether?

Lessons from the last hype cycle

Pittsburgh was an early hub of autonomous driving, made famous in 2015 by Uber’s high-profile poaching of 50 Carnegie Mellon University faculty. A driverless future seemed imminent. 

But the first generation of self-driving startups — from Uber ATG to Embark to Argo AI, another one-time Pittsburgh bright light — burned through billions chasing full autonomy in complex, urban environments. 

Trucking, with its long, relatively dangerous and at times monotonous trips that are vital to the American economy, seemed like a promising and necessary alternative to advance automated driving. The same challenging edge cases that cars face — construction zones, bad weather, unpredictable human behavior — slowed progress. 

While Uber closed the driverless trucking division it had acquired just two years earlier, Aurora’s approach became narrower and more disciplined. 

Aurora’s Texas operations are more like an automated train line than a fully independent car — albeit one where other drivers, animals and materials can suddenly enter the lane.

Its second route is a 600-mile stretch of I-10 that’s long, dry and predictable, ideal for controlled autonomy. By fixing its routes and mapping them to exacting standards, Aurora simplifies a still-daunting technical challenge. Back in September, I watched a demo of the company’s surveillance kit in Bozeman, Montana, where Aurora acquired a lidar startup and opened a satellite office.

It’s a clever pilot. But that could also be the ceiling. If driverless trucks can only operate on highly controlled corridors, their economic impact could be real but contained — a kind of robotic relay race between human-driven hubs.

Aurora’s founder and CEO, Chris Urmson, once led Google’s self-driving project. When the company went public in 2021 through a SPAC merger, he promised to “put Aurora’s mark on the world’s roads.” Since then, Aurora’s stock has traded below its launch price, a reflection of how investors have cooled on autonomy even as they pile into the AI boom.

Yet Aurora remains one of Pittsburgh’s anchor robotics employers, a backer of the Pittsburgh Robotics Network, and a visible example of how the region’s AI and autonomy ecosystem is maturing, even after the closure of local peers like Argo AI. Aurora acquired Uber’s ATG, and the Argo AI founders rebooted with trucking-focused Stack AV.

All told, boosters say this generation learned from the last cycle and is better because of it.

What happens to the drivers?

Today’s automation conversations rest on two separate questions: whether a given technology will advance as you expect, and even if it does, will it affect employment as you expect. 

A decade ago, the technology didn’t break through to wider adoption, and we’re still determining whether it will this time, either. Truck driving remains one of the most common jobs in America. The so-called driver shortage is really a retention crisis — grueling work, long stretches away from home and a pay system that rewards miles, not hours. Automate away the worst of these jobs, the boosters say.

What happens to the country’s 2.3 million tractor-trailer drivers?

Even if Aurora’s technology scales, regulatory approval, insurance frameworks and public perception will take years to catch up. Long-haul trucking may become semi-automated, but “last-mile” delivery — where a person hands you a package or backs into your loading dock — may stay human for a long time. There are several stages in trucking.

Standing on the convention floor in Pittsburgh, circling Aurora’s display truck, I was reminded of one of my guiding philosophies of reporting on technology: It’s easier to guess that something will happen than when that something will happen.

Driverless freight may not have arrived when the forecasts promised, but it may still arrive. The only question is whether we’ll be ready — not just with better sensors and software, but with an economy and workforce that can adapt as the wheel finally starts turning itself.