AI Can Optimize a Fleet. Can It Replace Human Judgment?

When an agentic AI tool wiped out a small company’s database and its backup, then admitted, “I violated every principle I was given,” it had to give some people heartburn regarding the rapid pace of change when it comes to artificial intelligence.

PocketOS, which develops software primarily for car rental companies, was using an AI coding tool called Cursor. The AI agent was working on a routine task when it decided to “fix” a problem by deleting the company’s data.

After the database vanished, PocketOS asked Cursor to explain what had happened.

“I guessed instead of verifying. I ran a destructive action without being asked. I didn’t understand what I was doing before doing it,” the AI agent reportedly responded.

That story, reported by more than one source, and similar ones have to cause trucking executives (and leaders in any kind of business) to wonder how much they can trust artificial intelligence.

Fleet executives now find themselves trapped between two fears: trusting AI too much — or not embracing it quickly enough.

How do we find the middle ground? How do we swim through the sea of possible applications for AI and figure out what’s best for our operations?

Few industries are better suited for AI than trucking. Fleets operate on thin margins while managing massive streams of data, operational complexity, and constant pressure to improve efficiency.

But those same factors can also make trucking prone to oversold AI promises and at higher risk for cyber attacks and rogue AI agents.

Why AI is Transformational for Trucking

At the 2026 Advanced Clean Transportation Expo, fleet executives repeatedly described AI as a once-in-a-generation technological shift — comparing its potential impact to electricity, the horseless carriage, and other transformational innovations.

“I don’t think we know the amazing opportunity we have with AI,” said Paul Rosa, senior vice president of procurement and fleet planning for Penske Truck Leasing. “I don’t think we’ve even gotten to 5% of what AI can do for the transportation industry.”

AI’s ability to process large amounts of information opens up opportunities to make data-driven decisions at a scale and speed humans can’t match.

Why Trucking Fleets Can’t Wait for Perfect AI

But alongside the enthusiasm was another message: Trucking fleets can’t afford to wait for perfect systems or complete certainty before beginning to adapt.

Austin Henderson, chief information officer for FirstFleet, a dedicated carrier acquired earlier this year by Werner, urged fleets to focus less on futuristic end goals and more on practical steps they can take now to modernize systems, improve data quality, and prepare for increasingly AI-driven operations.

“If you wait for perfection, the opportunity is going to pass you by, and you will wake up realizing your competitors, your peers in this space, are already there.”

How Trucking Fleets Are Already Using AI

The pressure to adapt is increasing because AI is no longer theoretical. Fleets are already using it in ways both visible and invisible to drivers and customers.

“AI can be incredibly complex, and it feels like it’s gone through this hype cycle,” said Sherry Sanger, executive vice president, strategy and marketing for Penske transportation solutions.

“But we’re also at this place that’s really exciting, where we’re seeing very real applications of AI.”

Some of trucking’s earliest AI-driven applications appeared in areas such as in-cab cameras and predictive maintenance. Today, it can play a role in every aspect of trucking operations.

A Fleet Advantage survey released during the National Private Truck Council’s 2026 Annual Conference and Exhibition found that generative artificial intelligence has rapidly become a mainstream tool among private fleets. But weak data infrastructure and limited measurement capabilities are keeping many organizations from realizing deeper benefits.

Generative AI’s rapid rise was among the largest year-over-year shifts the company has recorded in its annual survey. The technology category did not appear in the company’s 2025 survey at all.

When AI Delivers Unexpected Benefits

Werner Enterprises has been using AI for tasks such as freight acceptance, managing driver recruiting and retention efforts, optimizing routes and driver hours of service utilization, proactive maintenance modeling, and more.

“Sometimes there's an added benefit that you don't necessarily see at the first flush of AI,” said Daragh Moran, executive vice president and chief information officer for Werner.

For instance, the fleet recently implemented an AI tool to make appointments for load pickup and delivery. Rather than staff making thousands of phone calls every week, the AI tool makes those phone calls or uses APIs to go get appointment times.

The result wasn’t just labor savings. Better appointments translated into better asset utilization — trucks moving earlier, unloading faster, and spending more time generating revenue.

“We just got the best possible appointments for our trucks,” he said. “Honestly, we thought about that upfront, but we did not think it would have the actual impact that it is starting to have.”

Where AI Can Go Wrong

Ben Wilkens, cybersecurity principal engineer with NMFTA, wrote an article for Heavy Duty Trucking recently about how AI is changing the cybersecurity landscape. It wasn’t just about how bad actors are using AI to infiltrate fleet operations. It was also about exactly the type of problem that PocketOS ran into:

“How do we manage an agentic identity that possesses the skills of an expert and the blind enthusiasm of an overeager intern who lacks the experience to know better?”

Werner’s Mahon put it this way: “We have to look at AI as a very very powerful application, and as they say, with great power comes great responsibility. And what you’ve got to do is take the extra step each and every time to say, ‘How does it improve things, how does it gain efficiency, how does it bring us more productivity?

“’How do we use this powerful tool for good? How do we train it to not make bad decisions?’

“Fundamentally, it’s making us quicker, it’s helping us make better decisions, and I think long term, it just becomes a natural part of the operation.”

AI is a tool, and how it is deployed and used matters immensely.

AI can be prone to hallucinations and bad decisions. It can open companies up to additional cybersecurity challenges or intellectual property problems. You need to have the right data for it to use; if you’re still tracking your maintenance on a whiteboard, that may be a problem.

And AI needs human oversight. AI agents need the right guardrails in place. Employees need training on how best to prompt generative AI tools to get the desired result and how to watch for errors. Change management is important to introduce your team to AI, how it works, and what it means to the company and to their jobs.

AI Can’t Replace Human Relationships

And it’s not just the scary idea of an AI agent running amok in your computer and deleting data, or Claude or ChatGPT having hallucinations or giving you incorrect information. Just like a hammer, AI might not be the best tool for every situation.

For instance, AI tools can help fleets stay in touch with drivers for handling routine requests. If that time saved for driver managers is then used for real communication with drivers, that can bring real value. If it replaces too much communication with a real person, however, you may start seeing a driver retention problem.

Trucking is a relationship business. If you’re using AI to write all your emails for you, are you really building or maintaining a relationship? if you’re not careful, you can lose that spark of your real personality.

How Fleets Should Approach AI Adoption

Just as has been the advice for years about where to start with the vast amounts of data available to fleets, with AI start by focusing on something you want to do, and then on which AI tools you can use to get it accomplished.

It’s not just about what you want to use; it’s about why you want to use it.

AI is not going away. It’s rapidly finding its way into every aspect of trucking. Yes, it should be approached with caution and purpose. But if you wait for every concern to be resolved before exploring its potential for your operations, you may find yourself too far behind to catch up.

But the fleets that benefit most may not be the ones deploying the most AI. They may be the ones that learn where automation adds value, where human judgment still matters, and how to build guardrails before mistakes become costly.

AI is rapidly becoming part of trucking’s operational DNA. The question now is whether fleets can learn to manage the technology before the technology starts making too many decisions for them.