Trucking is an essential part of the American economy. When it comes to the physical weight of stuff moving around the country, more than 72% of the nation’s freight is carried on trucks, according to the American Trucking Associations.

But the job can be dangerous. It involves long hours alone on the road and fighting weather, exhaustion and other drivers who don’t think about truck-sized blind spots.

And now, trucking companies are adding more technology to the vehicles, including artificial intelligence. In theory, this kind of tech could act as another layer of safety by helping drivers stay awake or in their lane. However, crash numbers have actually risen since electronic logging devices became mandatory in trucks.

Karen Levy is a Cornell professor and author of the new book, “Data Driven: Truckers, Technology, and the New Workplace Surveillance.” She joined Marketplace’s Kimberly Adams to talk about what integrating AI into trucking means for truckers. A transcript of their conversation is below and has been edited for clarity.

Kimberly Adams: So when writing about AI and surveillance and technology, why choose trucking?

Karen Levy: Yeah, trucking is a really interesting industry through which to kind of think about and understand the way AI and surveillance are affecting work. So truckers are really the backbone of the economy, but we tend not to think about them very much. And I think not very many people have probably thought about the ways in which new digital technologies and AI are starting to affect trucking as a workplace. So that’s why I wrote “Data Driven.”

Adams: You say in your writing that if and when AI comes more fully into the trucking world, it won’t just completely replace drivers. So what will it do instead?

Levy: So autonomous vehicle technology, while it’s certainly advanced quite a bit over the past few years, is nowhere close to the point where it’s actually able to replace human workers. Instead, what we’re seeing is that AI is being felt in trucking by actually sort of invading the cab along with the trucker. So, among the things that are becoming more common in the industry are things like a camera that’s trained on a driver’s face that uses AI to monitor the flutter of his eyelids or how often his head is nodding. And the thinking there is that, you know, fatigue is a big problem in the trucking industry. And by using AI and camera systems, or biometrics, or wearables — that’s a better way to potentially manage drivers and to make them operate more efficiently in the vehicle. Of course, what we actually see is that a lot of those technologies end up driving out the very safest workers and really makes the job much less tenable. So, the narrative that we see is not really people getting kicked out of the cab, it’s that AI is really impinging on the dignity of the job.

Adams: Yeah, I was just wondering what the truckers think about all this.

Levy: Yeah, they’re not big fans. I mean, if you talk to truckers, a lot of truckers get into trucking because they don’t want someone looking over their shoulder all the time. Like, it’s a really common refrain among professional drivers that, you know, they liked the human autonomy that they were able to have. And a lot of new technologies that truckers are facing really are kind of an affront to that self-knowledge or that road knowledge that truckers have built up over years or even decades.

Adams: So are there short-term fixes that should be on the table to address some of these concerns that drivers actually have?

Levy: Yeah, so drivers for decades have noted that one of the biggest problems in the trucking industry is just under compensation. In 1980, truckers made about $110,000 a year in today’s dollars, and right now they make about $47,000 annually. And that number has been stagnant for many years. It’s really declined. And so what I think should happen and what truckers argue should happen, is kind of changing those root causes of overwork, right? Paying people for the labor that they’re actually doing, which would involve maybe changing the pay structure of trucking so that they’re not paid by the mile, but are paid actually for the number of hours that they work, and addressing other organizational issues in the industry to reduce the kind of fatigue that the technology is just being used to manage. Those types of solutions, they’re economic, but the problems in trucking really are economic problems. They’re not technological problems. And so trying to address them via surveillance technology, I think, will always be an inherently limited solution — and even a counterproductive solution. And it also drives out some of the safest workers.

Adams: What does the rollout of all this technology in the trucking industry mean for consumers?

Levy: I mean, one of the things that has been used to sort of justify some of these technologies is that consumers want safe highways. Like, I certainly want safe highways, and I think you want a safe highway. Nobody wants to be next to a trucker who’s been on the road for too long. And so these safety technologies are sometimes kind of justified on that basis. Unfortunately, what we know is that the electronic logging device and some of the other technologies that are used in trucking don’t have measurable impacts on safety. They actually don’t make the roads safer. If anything, the evidence we have suggests that they actually make the roads less safe. Crash rates have gone up since these things became mandatory. The other thing that people sometimes talk about is whether cost would go up for shipping, or, you know, would we have to wait longer for our Amazon Prime deliveries or something if we were actually to pay truckers for the work that they do? And the answer might be yes, but maybe those are the real social costs of depending on truckers’ essential labor to do all of these things. So, I think it just involves kind of a hard conversation about how much we’re willing to pay to actually have safe roads and to have essential workers work with dignity and be paid the wages that they deserve for that labor.