FAA Tests AI Traffic Tool In DC Before Wider Airspace Rollout

The Federal Aviation Administration is preparing to put an AI traffic-management tool into limited use over Washington, DC, before deciding how far it can move into the national airspace system, Ars Technica reported.

The SMART system is expected to advise controllers and traffic managers around the region’s three major airports as soon as September 21.

The first deployment would cover a congested slice of airspace before the agency considers expanding SMART into the full US system, which spans 29 million square miles.

SMART is designed to forecast traffic flows and identify possible conflicts from airline schedules, weather, airport capacity and airspace conditions.

The forecasts would put controllers, carriers and other aircraft operators on the same traffic picture, giving them a common basis for routing and departure-time decisions during congestion or weather recovery.

The limited start is a safety boundary as much as a launch plan.

Philip Mann, a principal consultant at Vector Strategic Consulting and a former FAA official, described the narrower scope as the right call because a national-scale AI system carries more unknowns than any single prediction.

In his view, reducing the operating area reduces the number of unknowns the agency must evaluate at once.

Airlines had been uncertain for weeks about what the agency intended to change.

Their concerns eased after the FAA clarified that SMART would not alter procedures for controllers or airlines and would instead provide alternative route information through existing FAA systems.

The system sits inside a broader $875 million, 12-year contract awarded in June to Boston-based Air Space Intelligence.

The award also covers a new Flow Management Data and Services platform for the Virginia command center that coordinates FAA traffic-control operations.

Mann has described the replacement flow-management system as the backbone and SMART as the predictive layer above it.

Air Space Intelligence’s separate Flyways AI product already manages more than 40 percent of US air traffic through airline customers including Alaska Airlines.

The company says Flyways uses a 4D digital twin of US national airspace to predict air traffic and weather conditions, but the source did not specify which AI model types will power SMART.

That distinction matters because deterministic AI, machine-learning models and generative AI behave differently.

Rule-based systems can return consistent results, while machine-learning systems infer patterns from probabilities.

Generative systems can vary their output and introduce accuracy risks, making model disclosure material for a safety-critical deployment.

Mann is also watching whether SMART remains limited to aircraft at 24,000 feet and above, the altitude band the FAA described in June.

That would place the software around cruise traffic and the climbs and descents feeding the Washington-area airports rather than every movement near the ground.

The first test comes while the FAA is modernising aging control infrastructure and managing a strained controller workforce.

The agency has begun replacing hundreds of radar systems dating to the 1980s, along with radios, telecommunications lines and voice switches that connect controllers with pilots.

Staffing pressure raises the stakes for any automation aid.

A Government Accountability Office report found the air traffic controller workforce declined 6 percent over the past decade, and shutdowns in 2013 and 2018–2019 froze hiring and training.

A 43-day government shutdown in 2025 put controllers on unpaid work, while FAA leaders later reduced the estimated controller requirement for 2026 to 2028 by about 2,000 positions.

The Washington deployment will therefore test more than routing efficiency.

The next expansion depends on how SMART performs under real workloads, degraded data and the unresolved question Mann identified: who owns the outcome when an AI-assisted prediction is wrong.