According to a report from McKinsey and Company, up to 90% of trucking companies use AI in their operations, but that could mean anything from playing with ChatGPT to deploying AI-powered dash cameras from a vendor. It doesn’t necessarily mean trucking companies are adopting AI as part of their strategy, but instead that they have tools they are experimenting with.
Mark Murrell, president of the driver training platform CarriersEdge, said many trucking companies work with vendors who are experimenting with AI and may even be incorporating AI into their back-office processes. But when it comes to safety, those stats are much lower.
Murrell said that mostly, only companies with around 1,000 trucks or more are experimenting with AI in safety—and even AI in general.
“If you're a large organization, and it is within your business strategy to be at the front of the pack on technology adoption, and you have the resources to be able to test things and do some experiments—and to suffer the failure of some of those experiments, because most of them are not going to work the first time—you absolutely should be doing something on that,” he said in a recent CarriersEdge webinar about avoiding the pitfalls of AI in safety.
He said maybe around 70% of large fleets in the CarriersEdge Best Fleets to Drive For program are not adopting but are “playing around with” AI. It doesn’t make sense for companies with fewer trucks than that to explore AI until it has been further proven because they wouldn’t see any benefit.
[RELATED: First steps into AI: practical use cases for fleets]
And AI just isn’t quite there yet when applying it to safety in trucking.
Human Touch
Many trucking companies have implemented AI chatbots on their websites, but they are often inadequate. Murrell said that form of implementation has never worked with drivers.
“The nature of the trucking industry is that you have to be very careful about how you implement this, obviously, from a safety and liability perspective, No. 1,” he said. “But just as important is the fact that you can't put an AI into any place that gets in the way of the human connection with drivers and have success with it.”
He said fleets need to use AI in a way that enhances those personal relationships rather than minimizing them. He also emphasized the importance of using AI as a tool to enhance safety rather than replacing human oversight.
Murrell warned that any technology, including AI, can expose companies to liability if not properly managed and supervised.
Driver cameras and safety systems, for example, still need human involvement. Those solutions can identify rolling stops, driver drowsiness, pedestrians, cellphones in hand, and so much more. But they also can have false positives, which garner negative reactions from drivers.
CarriersEdge has been obtaining driver feedback from its Best Fleets to Drive For surveys and has gathered that drivers understand the reasoning for deploying AI, but their support is lacking.
“They get the idea that it helps with safety, and that's good. They're on board with that,” Murrell said. “They do get fed up with the number of things beeping in their cabs and all of the times that those things really are not meaningful.”
He said AI is not the best solution in safety because relationships with drivers can be fraught in the best of times.
“You don't want to make that worse by having it automatically come from some machine or some bot that very much sounds like it's not a person,” Murrell said. “They would much prefer to have that conversation with a human. Yes, it is going to take you longer. It is going to require more time, but it is going to provide a lot of other benefits as well.”
Let’s get it started
Murell said driver cameras have begun to be widely adopted, and the same will happen with AI. Driver cameras and telematics, he said, are great places for fleets to start implementing AI. Predictive tools are getting better but are still having challenges, he added.
The key is to start slow.
Murrell recommended that fleets look at their businesses and determine what challenges they think could best be solved by AI. Don’t do what other fleets are doing, he added, unless they’re experiencing the same challenges, because every fleet is different.
From there, evaluate options to address the specific problem identified.
He said one method that tends to work well is designating one person to start investigating where the fleet could use AI and exploring some solutions to get a sense of what could work, what wouldn’t work, and how it might disrupt existing processes and potentially create additional challenges.
“The risk is that it ends up making things worse,” Murrell said. “That you end up maybe solving one problem and creating three others.”
When identifying a vendor to launch a pilot with, he advised fleets to be choosy. Learn about the vendor’s background: if it’s just in technology or if it also has experience in trucking. He said it’s important that the vendor understand the trucking industry. He also highlighted the need to select vendors that are transparent about their data practices.
“People have to be cautious about it, and don't worry if you're not doing very much because the time will come, and it will become very obvious how to use it,” Murrell said. “It will be like the internet. In 1995 and 1996, not very many of us were on the internet, but by 2000, everybody was once we figured out how to do it. It'll be the same thing with AI. It'll get to a point where everybody's using it, and it's very clear how to make a success of it.”
Electric avenue
AI seems like it’s everywhere, especially when trucking professionals attend conferences and most of the sessions involve discussions around the topic.
But Murrell said that’s because many of the panelists work for the tech companies producing AI products, and those companies are under pressure from their funding sources (private equity firms, venture capital firms, angel investors, etc.) to incorporate AI into their strategies, even if it isn’t always applicable. Because they’re pushing AI, customers think they need to get in on it as well, or they may be left behind, he added.
That’s not the case, Murrell said, comparing AI to the early days of electricity.
“There's something exciting. There's something electric in the air, but people aren't exactly sure how it is all going to play out,” he said. “It's like people have electricity and they're trying to figure out what to do with tools, so they're connecting their wires to their hammers and saws and electrocuting themselves in the process.”
Eventually, AI will be useful without the pitfalls. In the meantime, he said there’s a chasm between the mainstream market that just wants an end product that works and the early adopters that are willing to invest and potentially lose time and money—or as he put it, “get electrocuted”—to figure out the technology.
For the former, “it's something to be sort of aware of, that it's coming; it's going to change things, but it isn't necessarily anything that you should be worrying about just yet,” Murrell said.
He suggested that AI will become transformative, but it is not yet fully realized. AI is currently riddled with false positives because it is still learning and collecting data, he said, adding that an AI model may only miss 5% of the time, but that 5% can be significant.
So, “if you're not doing much yet, don't worry about it,” he said.
