FAA Expands AI Flight Planning Nationwide, Denies Ground-Stop Link

The Federal Aviation Administration has put its $875 million AI-supported flight-planning system into use across the airspace managed by all 22 U.S. air traffic control centers. The nationwide expansion occurred on Nov. 10, three days before a communications problem caused a separate airline ground stop. Government officials and developer Air Space Intelligence said the new system was not connected to that disruption.

https://youtu.be/PYpzQk7cH_Q

The timing placed an important engineering boundary in public view: The Strategic Management of Airspace, Routes and Trajectories system, known as SMART, can generate routing recommendations, but it does not control aircraft. Specialists review its output, the FAA command center distributes recommendations, and human traffic managers and controllers decide whether to use them.

What the system does

SMART brings together about 200 data streams covering weather, flight paths, traffic flow and controller staffing. Its AI-supported engine uses those inputs to present a consolidated picture of where traffic is heading, how much demand the airspace can accommodate and where weather or congestion may create constraints.

A team near Washington analyzes flight schedules, forecasts and other information through the platform. Recommendations then move through the FAA command center to air traffic personnel for consideration. That architecture keeps operational authority with people, but it also means the system’s value depends on how accurately it processes data and how effectively its output is communicated, checked and understood.

The FAA says a Washington-area test helped reroute 180 flights, producing average savings of 10 minutes and 80 gallons of fuel per flight. The agency also cited a Tampa-to-Dallas flight that cut 42 minutes using the system’s recommendations. Those are FAA-reported test results, and the nationwide rollout will expose the system to a much broader range of traffic, weather and regional operating conditions.

One intended benefit is more precise management of weather-affected airspace. Conventional restrictions can close large rectangular areas even though storms are irregular and may leave usable corridors. SMART is designed to help specialists anticipate how those conditions will move, identify alternative routes and reopen restricted airspace sooner. If the recommendations remain accurate, that could reduce delays and fuel consumption without transferring separation responsibility to software.

Human control requires more than a final approval

Keeping a person as the formal decision-maker is only one part of safe human-automation integration. Controllers and traffic managers also need enough training to recognize the system’s operating limits, question implausible recommendations and understand what information shaped an output. A poorly understood advisory can increase workload rather than reduce it, particularly when weather and traffic conditions are changing quickly.

That concern is behind an Oct. 13 request from House Democrats for the Government Accountability Office to audit SMART. The lawmakers cited limited workforce input and training, along with questions about reliability, safety safeguards and the possibility of false AI-generated information entering safety-critical decision-making. The National Air Traffic Controllers Association represents about 14,000 controllers, including roughly 3,000 trainees, making training quality and interface consistency consequential at national scale.

Former FAA official Mike O’Donnell said SMART must remain “just a tool in the toolbox” and called for methods to verify and validate its output. Air Space Intelligence CEO Phillip Buckendorf described the allocation of duties similarly: the software performs the computation, while human operators accept or reject its recommendations.

That division aligns with the FAA’s broader Roadmap for Artificial Intelligence Safety Assurance, which says AI should be treated as an algorithm rather than a human participant. The roadmap emphasizes clear responsibility, incremental deployment, workforce readiness and conventional systems-safety and human-factors practices wherever they remain adequate.

SMART’s advisory role also limits what the $875 million deployment can fairly be expected to accomplish. It may help traffic managers use existing airspace more efficiently, but it does not add controllers, repair communications infrastructure or independently resolve every source of delay. Its benefits will reach passengers only when its recommendations are timely, operationally usable and accepted by trained personnel.

The requested audit now provides the clearest next test of the rollout. Nationwide operation establishes scale; it does not by itself establish reliability across every operating condition. The central measure will be whether the FAA can document performance, training and safeguards well enough for controllers to reject a bad recommendation as confidently as they accept a useful one.

More aerospace and engineering stories, right in your MSN feed.
Follow AMI on MSN

By Thomas Caldwell — AMI’s senior editor for mechanical and mobility engineering, covering vehicle electronics, systems integration, electrification, chassis systems, propulsion, and safety policy.

Leave a Reply