FAA expands AI advice to manage air traffic congestion across entire US airspace

The Federal Aviation Administration has widened its use of an AI system called SMART from a single regional deployment to the airspace handled by all 22 air route traffic control centers in the United States. The agency announced the expansion in a statement posted on November 10, following an initial rollout in the National Capital Region that began on September 21. SMART advises controllers rather than flying aircraft: it generates recommendations, and humans remain in the loop at every stage.
The FAA's public figures center on the Washington-area trial. According to the agency, SMART helped reroute 180 flights in that airspace, a change it says saved an average of 10 minutes and almost 80 gallons of fuel per flight. The FAA has framed the national rollout as an attempt to reproduce those results at scale, citing reduced delays, lower fuel consumption and a better travel experience as the goals. The system carries a reported price tag of $875 million.
Functionally, the FAA describes SMART as a cloud-based tool that fuses weather patterns, flight paths, traffic flow data and controller staffing metrics to forecast how traffic will develop. From those predictions it proposes scheduling adjustments, identifies what it considers optimal routes, and works out the best sequence for arriving aircraft. In practice that makes it a decision-support layer sitting above the existing traffic management workflow, not a replacement for it.
Guardrails are explicit. A team of FAA specialists based in Washington validates SMART's recommendations before they reach individual air traffic control facilities, and local controllers can accept or decline what they receive. The agency stated in its September 21 announcement that the system cannot substitute for controllers or take control of an aircraft. That architecture matters for anyone assessing the risk: the automation proposes, a centralized human group filters, and a second human group on the ground decides. Each layer adds latency and judgment, but also limits how far a bad prediction can travel.
Industry reaction is split. Delta CEO Ed Bastian and American Airlines CEO Robert Isom offered supportive statements in the earlier FAA announcement. The National Air Traffic Controllers Association, which represents more than 15,000 members, told Flying magazine that it was not consulted or involved in the design, testing or implementation of SMART and therefore cannot speak to its effect on the workforce. That gap between operator buy-in and the workforce that works with the tool day to day is a notable point of friction for a system whose usefulness depends on controllers trusting its output.
The rollout has also proceeded in stages beyond the capital region. Before the nationwide expansion, the FAA extended SMART coverage to air route traffic control centers in Atlanta and in Jacksonville and Miami, Florida, according to The Dallas Morning News. The agency gave that newspaper an example in which SMART recommended a shorter, faster route for a cargo flight inbound to Dallas-Fort Worth International Airport. The flight was not identified, and the FAA has not detailed how the system may have aided passenger traffic.
A few things are worth separating as this scales. The headline metrics come from one dense, heavily managed airspace over a limited period, and averages across 180 flights say little about performance during irregular operations such as storms or ground stops, which is where congestion advice is hardest. A likely trade-off is that centralized validation keeps quality high but may slow recommendations precisely when traffic is changing fastest. It is also unclear from the public material how recommendations are surfaced to controllers or how acceptance rates will be tracked, though the FAA's stated intent to cut delays and fuel burn gives it obvious metrics to report against.
Why it matters: Controllers, airline dispatchers and flight operations teams across all 22 centers now work alongside a system whose suggestions can alter routes and arrival order, and airlines have signaled support for that shift. For developers building in aviation or logistics, SMART is a live example of human-in-the-loop decision support at national scale, where validation layers and controller discretion are as much a part of the design as the model itself. Whether the Washington numbers hold up nationwide, and whether the controllers' union's concerns get addressed, are the questions that will determine how far this model of AI advice extends into safety-critical operations.
Based on reporting from the original publisher. Visit the source for full context and later updates.
Publisher excerpt
Airlines praised the AI tool, but the air traffic controllers’ union wasn’t involved.