The Trump administration is turning to artificial intelligence (AI) to address the perennial chaos of American air travel.
But skeptics argue that if AI can’t manage a fleet of snack machines without error, trusting it with the complex, high-stakes ballet of thousands of commercial aircraft is a potentially grave gamble.
Transportation Secretary Sean Duffy recently unveiled a multibillion-dollar initiative aimed at modernizing the nation’s aging air traffic control system, placing high-tech software at the center of the Federal Aviation Administration’s (FAA) strategy to eliminate delays before they even happen.
At the heart of this modernization is a program dubbed Strategic Management of Airspace Routing Trajectories (SMART). According to Duffy, the software is designed to merge airline schedules with FAA data to predict congestion up to 45 days in advance. By identifying potential bottlenecks weeks ahead of time, the system can suggest micro-adjustments — shifting flights by as little as five to 10 minutes—to prevent the domino effect of delays that currently plagues major hubs.
“This software will say, ‘Well, listen, we can see this 45 days out,'” Duffy told CBS News. “Let’s move some of those flights a little bit later… and we can resolve the issue.”
The price tag for this digital overhaul is steep. Congress has already allocated $12.5 billion for physical infrastructure such as replacing ancient copper wiring and transitioning towers from paper strips to digital displays, but Duffy is now requesting an additional $6 billion to $10 billion specifically for the AI build-out. The Department of Transportation (DOT) estimates the total cost of the modernization project could eventually exceed $31 billion.
To develop the SMART platform, the FAA has tapped three major private-sector competitors: Palantir, Thales SA, and Air Space Intelligence. Palantir recently confirmed its involvement, noting that its data analytics tools will be used to advance aviation safety and modernization objectives.
The prospect of “AI in the cockpit” of ground control has naturally raised concerns regarding safety and job security. However, Duffy has been adamant that the technology is a co-pilot, not a replacement. “Am I gonna replace a controller and have AI manage the airspace? The answer to that is hell no,” he said.
Nick Daniels, head of the National Air Traffic Controllers Association, echoed this sentiment, viewing the tools as a way to alleviate the “excessive workload” that has been blamed for several recent near-misses and fatal ground collisions. Justin Ciaccio of Peraton, the prime integrator for the project, suggested the goal is to make controllers “superhuman” by providing them with better data, rather than displacing the workforce.
Despite optimism, critics point to the checkered history of AI in critical infrastructure. Large language models and predictive algorithms have frequently struggled with hallucinations and failures in simpler tasks, such as staffing schedules and even automated vending systems.
“Using AI to modernize air traffic control is directionally right. The current system is long overdue for an upgrade. But the real test won’t be the technology itself, it’ll be integration,” Joseph Hoefer, chief AI officer at Monument Advocacy, said in an email. “The FAA is layering new AI tools onto decades-old infrastructure, and that’s where delays, cost overruns, and safety concerns tend to emerge.”
“The FAA’s challenge isn’t just building smarter software…it’s deploying it safely within one of the most complex, risk-averse systems in government,” he said.
The DOT aims to have the overhaul completed by the end of 2028, timing the project’s finish with the conclusion of the current administration. As the FAA begins the arduous process of developing, debugging, and training on these new systems, the flying public can only hope that these SMART trajectories lead to smoother landings.
