US Expands AI for Air Traffic Controllers: Flights Saved an Average of 10 Minutes and 300 Liters of Fuel
The US Federal Aviation Administration (FAA) is rolling out SMART, an AI assistant, to all 22 of the country's air traffic control centers. It suggests how controllers can reroute flights and reorder landing queues, taking weather and airspace congestion into account. According to the FAA, 180 flights over Washington had their routes changed, and each one saved an average of 10 minutes and more than 300 liters of fuel. These figures have so far come only from the agency itself, and the final decision still rests with the controller.
AI-processed from Habr AI; edited by Hamidun News
Air traffic controllers juggle a dozen variables at once every day: thunderstorms, congested stretches of sky, short-staffed shifts, and a queue of planes waiting to land. The Federal Aviation Administration (FAA), the American aviation regulator, is handing part of these calculations to SMART, an AI assistant, and expanding it to all 22 air traffic control centers in the country. While the system was running over Washington, its recommendations led to route changes for 180 flights. According to the FAA, each of them saved an average of 10 minutes and more than 300 liters of fuel. For practitioners, this is a live example of an AI advisor in a critical industry where the final say still belongs to a human.
What SMART does and how it works in practice
The system's full name is Strategic Management of Airspace, Routes, and Trajectories. It is a cloud service that cost $875 million and brings together weather data, flight trajectories, traffic flow, and air traffic control staffing in one place. Based on this data, it forecasts how traffic will look over the next few hours and issues recommendations: how to plan departures, which routes to choose, and the order in which to accept aircraft.
Imagine a navigation app that sees not only the jam right in front of you, but also that a column of trucks is about to stall at the next interchange and the dispatch center is short-staffed. It proactively suggests a detour. SMART does the same for airliners, only the stakes are higher.
An example from the FAA itself: for a cargo flight arriving at Dallas/Fort Worth Airport, the system proposed a shorter and faster route. However, as the Dallas Morning News pointed out, the agency did not name the specific flight and did not explain what SMART contributed to passenger operations.
Numbers and scaling
The first deployment took place over Washington on September 21, 2026. Since then, the routes of 180 flights have been changed based on the system's recommendations. The FAA claims that each one saved about 10 minutes and more than 300 liters of fuel on average. A rough calculation from these figures gives roughly 30 hours of flight time and at least 54,000 liters of fuel over the entire period of operation. For aviation, that is a modest amount, and the sample is small and limited to a single region.
The coverage was then extended to the control centers in Atlanta, Jacksonville, and Miami. SMART will now be connected to all 22 centers in the country. According to the agency, the goal is to reduce delays, cut fuel consumption, and improve passenger service. The rollout has been endorsed by Delta Air Lines CEO Ed Bastian and American Airlines President Robert Isom: for airlines, saved minutes and liters translate directly into money.
Who makes the decisions and what remains open
The FAA stresses that a human stays in the loop. Recommendations are reviewed by a team of FAA specialists in Washington before they are passed on to the control centers. Local controllers may accept the advice or ignore it. The agency also states that the AI system does not replace controllers and cannot take control of an aircraft.
Still, plenty of questions remain.
- All the figures on 10 minutes and 300 liters so far come from the regulator itself. There is no independent assessment, and it is unclear what the "saved" time was compared against.
- The National Air Traffic Controllers Association (NATCA), which represents 15,000 people, said it was involved neither in developing nor in testing the system. As a result, the association cannot judge how SMART affects its staff.
- Central review of recommendations in Washington could become a bottleneck once the system works nationwide. This is a conjecture, not a statement from the FAA.
- Ars Technica previously reported on unresolved questions surrounding SMART and on the controller shortage the FAA has been dealing with for years.
What practitioners should take away
First, a "model proposes, human confirms" scheme is easier for regulators and users to accept than full autonomy. For logistics, energy, or manufacturing, this is a workable template.
Second, the rollout path is typical and sensible: one region, then neighboring centers, then the whole country. But success is so far measured by two metrics, time and fuel, and both are reported without external audit. If you are launching a similar solution, agree in advance on the metrics and on who verifies them.
Third, ignoring the people who have to work with the system is costly. If a union learns about a project after the fact, trust in the recommendations drops, and manually rejecting them becomes the norm. SMART's real value will become measurable once independent data covers all 22 centers and it is known how often controllers agree with the suggestions.
- Nov 21, 2026FAA rolls out SMART AI advisor across US airspace: 10 minutes and 80 gallons of fuel saved per flight
- Nov 21, 2026FAA rolls out SMART AI advisor nationwide: 10 minutes and 80 gallons saved per flight
- Nov 21, 2026US Aviation Regulator Launches $875M AI System SMART: Test Flights Saved 10 Minutes Each
- Oct 29, 2026OES Puts 30 AI Agents on Mining Haul Trucks and Lifts Haulage by 10%
Key Questions Answered
What SMART does and how it works in practice?
The system's full name is Strategic Management of Airspace, Routes, and Trajectories. It is a cloud service that cost $875 million and brings together weather data, flight trajectories, traffic flow, and air traffic control staffing in one place. Based on this data, it forecasts how traffic will look over the next few hours and issues recommendations: how to plan departures, which routes to choose, and the order in which to accept aircraft.
What are the key details on Numbers and scaling?
The first deployment took place over Washington on September 21, 2026. Since then, the routes of 180 flights have been changed based on the system's recommendations. The FAA claims that each one saved about 10 minutes and more than 300 liters of fuel on average. A rough calculation from these figures gives roughly 30 hours of flight time and at least 54,000 liters of fuel over the entire period of operation. For aviation, that is a modest amount, and the sample is small and limited to a single region.
Who makes the decisions and what remains open?
The FAA stresses that a human stays in the loop. Recommendations are reviewed by a team of FAA specialists in Washington before they are passed on to the control centers. Local controllers may accept the advice or ignore it. The agency also states that the AI system does not replace controllers and cannot take control of an aircraft. Still, plenty of questions remain. All the figures on 10 minutes and 300 liters so far come from the regulator itself.
Ship AI Faster with Founders, Engineers & Weekly Architecture Reviews
Inside Vibe Club (hamidun.com/club) — weekly live masterminds with Zhemal Khamidun, Tech Radar, full LMS access, and 370+ production Claude Code skills.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.