US Aviation Regulator Launches $875M AI System SMART: Test Flights Saved 10 Minutes Each
The US Federal Aviation Administration (FAA) launched SMART nationwide on November 10, an AI system that cost $875 million. It combines 200 data streams on weather and traffic and advises controllers where congestion is likely to form soon. In an autumn test, 180 flights changed routes and saved an average of 10 minutes and 303 liters of fuel each. Humans make the decisions, and authorities do not link SMART to the November 13 outage. The figures so far come from the agency itself, and there has been no independent verification.
AI-processed from TechXplore AI; edited by Hamidun News
Imagine every air traffic controller in American aviation getting an assistant that spots, in advance, where a jam or a thunderstorm is about to form in the sky, and suggests going around the problem before it appears. That is how the FAA, the Federal Aviation Administration, describes SMART. The agency spent $875 million on the system and switched it on across the country on November 10.
In an autumn test near Washington, 180 flights changed routes, and each one saved an average of 10 minutes and 80 gallons of fuel (303 liters). Three days later, on Friday, November 13, American aviation came to a halt: the FAA stopped departures nationwide because of communication problems. The timing raised questions about what this system does, what risks it carries, and who is responsible for its mistakes.
For anyone deploying AI where the cost of failure is high, this is a live case.
What SMART is and what it does
SMART stands for Strategic Management of Airspace, Routes and Trajectories. It was developed by the startup Air Space Intelligence (ASI). The system runs from a single center near Washington. There, a team of specialists looks at flight schedules, weather forecasts, and hundreds of other data streams. Recommendations then go to the FAA's command center and from there reach controllers across the country.
According to the FAA, the platform integrates 200 data streams: weather, routes, traffic load, and data on controller staffing. The AI engine shows where planes are heading, how much traffic the sky can handle, and where a jam or bad weather is likely to emerge soon. The idea is for specialists to fix a problem before it happens.
Thunderstorms are a telling example. Today the FAA closes weather zones with large rectangles. Real storm fronts look different: they are ragged, and between them there can be narrow corridors where flying is safe. SMART is supposed to show how the weather will change over time. That would let controllers route around a storm and use open sky, rather than closing it off entirely.
Test figures and the limits of responsibility
In the autumn, the system was tested in the Washington area. According to the FAA, it helped reroute 180 flights. On average, each one saved 10 minutes and 303 liters of fuel. The best case was a flight from Tampa to Dallas that cut its route by 42 minutes. These are the agency's own figures, and there is no independent verification yet.
On November 10, SMART was connected to the airspace of all 22 air route traffic control centers. Officials and ASI share one position: no link has been found between the system and the halt in departures. Homeland Security Secretary Markwayne Mullin, the leadership of the Department of Transportation, and the FAA have all said so. ASI stated that the outage was not caused by SMART, by any of the company's products, or by its integration with the FAA and the traffic management service.
Importantly, SMART does not control flights. It issues recommendations, and a human decides whether to accept them. ASI's CEO, Phillip Bukendorf, told NPR that the AI counts and calculates, but the "decider and judge" is the operator.
Why the questions remain
Criticism emerged even before the outage. On October 13, House Democrats sent a letter to the GAO, Congress's audit office, asking for a review. The letter says that safety specialists took little part in development and received little training. The authors worry that the AI could hamper experienced controllers or produce fabricated information, which is critical in aviation.
Mike O'Donnell, a former FAA executive who spent 17 years at the agency, advises treating SMART as one tool in a kit, not as a machine that does the job. In his words, its conclusions need checking and validation, and trust should come only after the AI proves its reliability again and again. In his assessment, that will take time.
The context matters too. The industry has long been short of controllers: the NATCA union was warning of a crisis as early as 2015. It now represents about 14,000 controllers, roughly 3,000 of them trainees. Experts agree, however, that AI will not replace people who spend years training for this job. According to former FAA manager Michael McCormick, no working equivalent of SMART exists in any other country. Eurocontrol is testing a similar system, and several manufacturers are building prototypes.
What this means for practitioners
The lesson is not only about aviation. When AI only advises and a human decides, deployment is simpler, but that does not remove the questions. It requires training people, verifying recommendations, and having a clear plan for when the model gets things wrong. If the system's users did not take part in building it, they will trust it less. Until there is an independent audit, the savings figures remain the client's claims. The November 13 outage, by all indications, was not related to SMART, but it showed how quickly a hasty rollout of AI in critical infrastructure becomes a public debate.
- Nov 21, 2026US Expands AI for Air Traffic Controllers: Flights Saved an Average of 10 Minutes and 300 Liters of Fuel
- Nov 21, 2026FAA rolls out SMART AI advisor nationwide: 10 minutes and 80 gallons saved per flight
- 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 planner and cuts controller target from 14,633 to 12,563
Key Questions Answered
What SMART is and what it does?
SMART stands for Strategic Management of Airspace, Routes and Trajectories. It was developed by the startup Air Space Intelligence (ASI). The system runs from a single center near Washington. There, a team of specialists looks at flight schedules, weather forecasts, and hundreds of other data streams. Recommendations then go to the FAA's command center and from there reach controllers across the country.
What are the key details on Test figures and the limits of responsibility?
In the autumn, the system was tested in the Washington area. According to the FAA, it helped reroute 180 flights. On average, each one saved 10 minutes and 303 liters of fuel. The best case was a flight from Tampa to Dallas that cut its route by 42 minutes. These are the agency's own figures, and there is no independent verification yet. On November 10, SMART was connected to the airspace of all 22 air route traffic control centers.
Why the questions remain?
Criticism emerged even before the outage. On October 13, House Democrats sent a letter to the GAO, Congress's audit office, asking for a review. The letter says that safety specialists took little part in development and received little training. The authors worry that the AI could hamper experienced controllers or produce fabricated information, which is critical in aviation.
Test Frontier LLMs & Autonomous Agents in One Workspace
Run Claude Opus 5.5, GPT-6, Gemini 3.1, and DeepSeek side-by-side in Nomad AI (nomadnet.ai) — unified model routing, deep research, and multimodal workflows under one subscription.
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.