Are artificial safety nets setting you up for a fall?

Artificial intelligence is a powerful tool. But it’s not powerful enough yet to replace the human side of fleet safety. Don’t let this emerging technology create an artificial safety net that your fleet might regret.

Nothing in transportation is more important than safety. Every good fleet manager wants their drivers to return home safely from their routes. Every shipper wants their products delivered in full and on time. And every fleet owner doesn’t want to be on the wrong side of the next nuclear verdict because a prosecutor convinced a jury their fleet was unsafe.

Enter artificial intelligence. Technology providers are heralding it as the key to better operations. Machine learning and large language models are capable of analyzing data quickly to help fleets develop more effective routing and predictive maintenance strategies, thereby streamlining operations and improving the bottom line.

See also: Practical AI tools in transportation management

But this evolving technology also comes with risks. The most significant risk is putting too much faith in a machine to make your fleet safer.

“We live in a world where AI is merging with our day-to-day lives in every conceivable way, and the speed and depth of its integration into our lives in both scope and velocity is stunning,” according to Sean Ritchie, VP of sales and pre-sales at Solera Fleet Solutions.

Speaking at a recent Reuters U.S. transportation conference, Ritchie offered a stark warning to fleets and transportation technology providers: Putting too much faith in AI could expose companies to significant risks on roads and in courtrooms.

“I am concerned that in our rush to adopt AI, specifically within fleet safety, we have not given proper consideration to the implications of its adoption, given the reality of what that technology actually is today—not what it’s going to be in three to five years,” Ritchie explained.

He noted that even the best fleet safety systems on the market still trigger about 10% false positives, while most trigger even more. And if a system is finding false positives, it is likely not detecting actual unsafe actions. Ritchie said that about 20% of a fleet’s drivers tend to represent about 80% of an operation’s vehicle risks. To improve your fleet’s safety, you need to focus on that 20%.

“False positives complicate your ability to do so, potentially having your safety team spend valuable time focusing on drivers who are not as urgent a priority,” he said. “You simply can’t afford to waste 10% of your time.”

While a fleet manager might be diverted from focusing on the most problematic drivers on staff, AI can still overlook some actual driving risks. It might prioritize one bad action, such as rolling through a stop sign, while ignoring countless others.

Ritchie told me that Solera’s internal research revealed that AI-only video safety systems miss one-third of risks.

“You are liable for acting on all data that is presented to your team,” Ritchie said when talking about how AI ties in with nuclear verdicts. “If a video only has the AI trigger list, but there is another risk on that video that is not labeled and highlighted for coaching, you are still liable for coaching that other risk and documenting that you coach that risk.”

If a driver has an at-fault accident while caught on video doing something unsafe, an opposing counsel will use discovery to “subpoena every single video of that driver,” Ritchie said. “They will go through that video, frame by frame, and document everything risky that driver did. Your defense can’t be: ‘Well, my AI system didn’t label it for coaching, so I didn’t coach it.’”

What’s the answer to combating these AI limitations? It’s human intelligence. Until trucking perfects AI coaching and safety detection, fleets and drivers need human-to-human interactions and understanding.

“I believe that as long as we are honest with each other, put the safety of the drivers as the most important thing in our business, and ultimately find the right balance of artificial intelligence and human intelligence, we will find operational optimizations that will protect our drivers, our fleets, and our businesses,” Ritchie said.

While your fleet focuses on how AI can give you an edge, don’t forget the human aspect of the industry.