America is grappling with an air traffic controller shortage. Can AI help?

By Lisa Eadicicco, CNN

(CNN) — The longest government shutdown in US history shook travel plans across America earlier this month as a shortage of air traffic controllers resulted in thousands of delayed or canceled flights.

But even the end of the shutdown doesn’t fix the the broader air traffic controller shortage that’s been plaguing American flyers for years. Artificial intelligence, with its ability to analyze troves of data and potentially upend industries, might be able to help. After all, some in the field have called AI as impactful on everyday life as electricity or fire.

While AI does have some practical uses in aviation like crunching data on weather patterns or aircraft trajectories, experts highlighted a range of reasons why the technology won’t be a substitute for air traffic controllers anytime soon – or likely ever.

“It’s a human system, and it’s human-centered for sure because it demands almost near perfect performance,” said Ross Sagun, a former pilot and air traffic controller with more than 50 years of flight experience. “The smallest mistakes can turn into giant disasters.”

Why AI won’t replace traffic controllers

Air traffic controllers play a critical role in aviation, guiding jets safely both on the ground and in the air, ensuring safe distances are maintained between aircraft and communicating with pilots – skills that require multitasking, quick decision-making and intense concentration. The Federal Aviation Administration says it handles more than 44,000 flights a day, and 5,500 are in the sky at peak operational times.

Nick Daniels, president of the National Air Traffic Controllers Association, described the job as demanding “unique skill and precision” in his written testimony for a hearing in March before the US House of Representatives’ Subcommittee on Aviation of the Committee on Transportation & Infrastructure. He called air traffic controllers the “backbone” of the National Airspace System, adding that they “require rigorous training, a mastery of complex systems, and the ability to perform under immense pressure.”

That pressure is exactly why the job can’t be outsourced to AI, according to Sagun. Hearing emotion and stress in a pilot or fellow controller’s voice is a key signal that something is wrong and it’s unclear whether AI can detect those cues.

“That’s where you jump in, you scream for help, and we get people over there to take over the situation,” he said.

The rate at which AI is evolving and changing would also make it hard to certify the technology, he added.

There’s little to no margin for error in air traffic control given the safety risks. Even if the technology were to be used for automating the process of guiding aircraft as they land or take off, human workers would still need to manage unforeseen circumstances, according to Amar Gupta, a research scientist with MIT’s Computer Science and Artificial Intelligence Laboratory.

And some research as far back as the early 2000s suggests that type of arrangement, in which a human worker intervenes in the event of an emergency or automation error, may not always be effective. One study from 2005, for example, found that controllers were more likely to detect incidents manually than when using an automated tool that wasn’t 100% reliable.

Air traffic controllers might not be as vigilant if automation were to play a bigger role in their jobs since they wouldn’t be as actively engaged in the environment, according to Michael McCormick, associate professor and program coordinator for air traffic management at Embry-Riddle Aeronautical University.

“That is the one thing that humans are very poor at doing — that very quickly, you lose vigilance, generally through boredom,” McCormick said.

How AI could be helpful for air traffic controllers

Experts say there is a place for AI to help with air traffic control, but it would just be to assist with duties rather than replacing jobs.

With many air traffic controllers either retiring or finding other jobs during the shutdown, such help may be welcome. Air traffic controller staffing has plummeted since 2012, according to data cited by the National Air Traffic Controllers Association (NATCA), and the Federal Aviation Administration’s Air Traffic Controller Workforce Plan indicates the number of certified professional controllers fell below targets in 2024.

AI could be useful for helping controllers understand weather patterns, potentially helping to predict the possibility of storms, according to Sagun.

The FAA is working toward an upgraded system that uses AI to help analyze trajectories from aircraft in flight, estimate possible conflicts and provide recommendations to controllers, according to McCormick, though he says it is still years from testing and implementation.

Searidge Technologies, owned by United Kingdom-based air traffic control services provider NATS, has developed an AI system for monitoring ground movements and facilitating communication with pilots.

The technology is being used at London’s Heathrow Airport to track aircraft exiting the runway and Florida’s Fort Lauderdale-Hollywood International Airport to monitor gate occupancy.

The company says on its website that the technology isn’t a replacement for human control. It’s seeing “strong global interest” in its system and expects more “operational rollout and demonstration sites in the coming year,” Ron Reddick, vice president of global sales for Searidge Technologies, said in an emailed statement to CNN.

But don’t expect major upgrades to happen overnight in the United States.

Implementing new technologies into air traffic control requires rigorous testing and is a lengthy ordeal, backed by infrastructure designed to last for decades. The Department of Transportation plans to upgrade the FAA’s air traffic control system by 2028 with “all new hardware” and “all new software,” Secretary Sean Duffy said in May. But those upgrades are now going to take longer than initially planned, Duffy recently told reporters at a press conference.

“The entire process takes years,” said McCormick when asked how long it would take to implement AI in current air traffic control systems. “We’re in initial stages. We’re not far down the road.”

The-CNN-Wire
™ & © 2025 Cable News Network, Inc., a Warner Bros. Discovery Company. All rights reserved.