US officials say an $875 million AI-powered air traffic management system played no part in Friday’s nationwide flight ground stop — but its rollout days earlier has intensified questions about relying on AI in safety-critical systems.
The Federal Aviation Administration’s SMART system (Strategic Management of Airspace, Routes and Trajectories) went nationwide on 10 November, three days before communications failures grounded flights across the US. Officials and the system’s developer, startup Air Space Intelligence (ASI), say the two are unrelated.
What SMART does
SMART combines around 200 data streams — weather, flight paths, traffic flow and controller staffing — to predict congestion and suggest alternative routes. A specialist team near Washington reviews its output and sends recommendations to the FAA’s command centre, which passes them to air traffic controllers.
It does not control flights. “The AI does the math, the AI does the computation,” ASI chief executive Phillip Buckendorf told NPR, but human operators decide whether to act. “They are the decider, they are the judge.”
In an initial test around Washington, the FAA says SMART helped reroute 180 flights, saving an average of 10 minutes and 80 gallons of fuel each. No other country has a working AI air traffic tool of this kind, according to Embry-Riddle Aeronautical University’s Michael McCormick; Eurocontrol is testing one.
The concerns
In October, House Democrats asked the Government Accountability Office to audit SMART, warning that frontline aviation staff had little input into its development and limited training, and raising the risk that AI could “hallucinate safety-critical information.”
Former FAA official Mike O’Donnell said the agency “cannot afford to be wrong,” and that AI must remain “just a tool in the toolbox” with clear ways to verify what it produces.
The rollout comes amid a long-running shortage of US air traffic controllers. Experts say no AI system can replace them.





