Air traffic control is back in the news. On Apr. 28 and again on May 9, communication system breakdowns affecting Newark Liberty International Airport limited the ability of air traffic controllers to guide and manage the dozens of airplanes in the skies above them — for as long as 90 seconds.
United Airlines CEO Scott Kirby expressed deep concern for his company’s customers, asking for a resolution to air traffic control problems at the airport that have persisted for years. United also announced cuts to its flight schedule out of Newark, which serves as one of its hub airports.
Air traffic control relies on both radar and satellite technologies to track airplanes in the sky. Radio communication and radar technology were introduced into air traffic control back in the 1930s and 1940s, respectively. They continue to be integral components of how air traffic controllers do their job. Yet the modern era’s volume of flights is overwhelming how controllers communicate with flight crews and perform flight tracking.
This problem was already recognized two decades ago. It prompted Congress to pass the Vision 100 – Century of Aviation Authorization Act of 2003, launching the NextGen initiative to modernize the National Airspace System. The FAA Modernization and Reform Act of 2012 and the FAA Reauthorization Act of 2024 further fueled NextGen air traffic control improvements.
A recent proposal by the FAA to build a new air traffic control system suggests major changes are coming, all of which sound encouraging, though they will take several years to be implemented.
Despite all past efforts, system failures occur. What happened at Newark Liberty International Airport was less about a manpower shortage and more about a system breakdown. The after-effects of air traffic controllers taking trauma leave only exacerbated whatever manpower shortages the airport already had.
The Federal Aviation Administration is responsible for hiring and training air traffic controllers. Given that they oversee 45,000 flights each day on average in the U.S., with around 27,000 of them commercial — that is, 60 percent — their job is as critical to safe air travel as that of those who fly and maintain the airplanes.
Yet finding such people and filling positions has become exceedingly difficult. This has pushed the FAA to use incentives to retain retirement-eligible air traffic controllers and ramp up recruitment efforts targeting former military. Such programs help solve the long-term plans to staff control towers. They do little, however, to resolve short-term shortages.
The FAA claims that there is an immediate shortage of around 3,000 air traffic controllers, with over 14,000 on the job as of September 2024. The criteria to become an air traffic controller are stringent; Applicants must be U.S. citizens, under 31 years old, and meet medical and security thresholds. The bar for acceptance is sufficiently high that just 10 percent of all applicants meet the standard and enter the training program. Given how critical air traffic controllers are to the safety of the National Airspace System, and the stressful environment under which they must work, relaxing any such criteria should not be an option.
If the air traffic control system must continue to provide the highest level of performance, simply adding more staff to the existing model is not the solution. What happened at Newark would not have been averted with more air traffic controllers. What is needed is an overhaul of the air traffic control system and the technology being used.
Given that air traffic controllers manage airplanes across the entire National Airspace System, from when they leave their gate to moving to the runway for takeoff to their flight and eventual landing, and then to their gate at their destination, the multitude of tasks can be daunting.
This is how artificial intelligence can play a role. AI cannot replace air traffic controllers, nor should it. AI can, however, support their efforts, making it possible for air traffic controllers to be more efficient and effective, and perhaps reduce the risk of miscommunications that raise the risk of runway incursions and other incidents.
The number of runway incursions in fiscal 2024 was 1,758, or around 146 per month. This monthly rate has been mostly flat since 2019, with 2020 and 2021 showing lower rates — aberrations, given the impact of the COVID pandemic on air travel. Most runway incursions are benign, with airplanes out of a well-defined position. The risk associated with such incidents is exceedingly small.
The riskiest runway incursions, labeled Category A and B, occurred just nine times in 2024. Given that they occur at such a low rate and on the ground at airports, the potential to reduce such runway incursions with AI should not be ignored. Moreover, the typically mundane operations into and out of gates can also be supported with AI.
Adding more air traffic controllers is not a bad thing; designing a long-term plan for recruiting and retaining them is much needed. However, if the very system that they are overseeing remains the same, what would be just as important is a plan for upgrading the technology and incorporating more AI into air traffic control operations.
As our nation’s airways become more crowded, and as more flights are needed to fill an ever-growing demand for air travel, the management of the National Airspace System must also adapt. Hoping to simply scale air traffic control operations with more of the same is certain to hit a roadblock. What occurred in Newark could occur at any airport, at any time. Given how the hub-and-spoke system employed by the largest airlines interconnects all airports in the airlines’ networks, a disruption at Atlanta Hartsfield, Dallas Fort Worth International or Chicago O’Hare could affect hundreds of airports around the nation.
Now is the time to rethink how AI can become an invaluable support for air traffic control.
Sheldon H. Jacobson, Ph.D., is a computer science professor in the Grainger College of Engineering at the University of Illinois Urbana-Champaign. A data scientist, he uses his expertise in risk-based analytics to address problems in public policy.
