Free Intensive Nov 21–22Free 2-day hands-on Vibecoding Intensive on Nov 21–22 at 16:00 MSK (Sat–Sun) · Live Cohort 3 starts Nov 23Join Free →Cohort 3 (Nov 23)
ReleaseInsight · On RadarWired→ original

FAA rolls out SMART AI planner and cuts controller target from 14,633 to 12,563

The FAA has been using the SMART AI program since September. It shows air traffic controllers ahead of time where aircraft trajectories will intersect and where traffic will stall. In May, the agency lowered its target for active controllers from 14,633 to 12,563, citing new planning tools. After the nationwide ground stop on October 13, authorities said SMART was not linked to the communications outage. The takeaway: an AI recommendation in a critical system is useful, as long as a human can still spot its mistakes.

AI-processed from Wired; edited by Hamidun News
FAA rolls out SMART AI planner and cuts controller target from 14,633 to 12,563
Source: Wired. Collage: Hamidun News.
Listen to article
Article Fact Sheet & Key Figures
ParameterDetails
Primary sourceWired ↗
Editor-in-Chief
Key metric #1Joby flew a Cessna Caravan 3,200 miles without touching the controls, with a pilot on board
Amazon500 drone delivery sites by year-end; Walmart: 270 by the end of 2027
Key metric #3Democrats in Congress have asked the Government Accountability Office to check SMART for "hallucinations"

Imagine a navigation app that doesn't just show traffic jams but predicts, an hour ahead, where dozens of cars will get stuck at once, and tells you to take a detour. That's roughly the kind of tool the US Federal Aviation Administration has rolled out across the entire national airspace. It's called SMART, and on September 27, US Transportation Secretary Sean Duffy presented it in a video on X captioned "GAME CHANGER."

Almost three weeks later, on Friday, October 13, the FAA announced a nationwide ground stop, and flight schedules around the world came apart. Authorities quickly said there was no cyberattack and that SMART had nothing to do with the communications failure that prompted the halt. But when Homeland Security Secretary Markwayne Mullin was asked whether an AI error could be ruled out definitively, he would only say: "Nothing is certain right now."

What SMART does and who is behind it

The name stands for Strategic Management of Airspace, Routes and Trajectories. The program was developed by Boston-based Air Space Intelligence, and the FAA announced its deployment in September. It is predictive AI: it processes large volumes of flight data to warn in advance about scheduling problems and airspace congestion. FAA Administrator Bryan Bedford says that, together with equipment upgrades, SMART will help reduce delays, optimize traffic, and keep capacity open without compromising safety.

According to the authorities, the system is not supposed to replace controllers: it makes suggestions, and people make the decisions. Michael McCormick, a professor of air traffic management at Embry-Riddle Aeronautical University, sees an advantage in the tool: when a computer can see an aircraft's entire trajectory, conflicts can be spotted early. Aviation attorney and former US Department of Transportation Inspector General Mary Schiavo considers SMART an important improvement and points to what the current environment risks: close aircraft encounters, runway incursions, delays, and ground holds.

Two risks: model error and fewer people

The first risk will be familiar to anyone who has worked with language models. If the program produces nonsense, will the operator notice? About two dozen Democrats in Congress reportedly wrote to the US Government Accountability Office asking it to review SMART. They fear the system could "hallucinate safety-critical information."

The second risk concerns staffing. The US has a chronic shortage of controllers, and the FAA hopes AI will reduce the need for people. In May, the agency lowered its target for active controllers from 14,633 (the 2024 goal) to 12,563, citing "modern staffing models and planning tools." Critics worry that the promised efficiency will be used as a reason to cut staff, even though savings from AI have not yet materialized in other industries. A recently retired FAA employee, who asked to remain anonymous, called the situation a "Pandora's box" and accused authorities of deploying technology without understanding the complexity of the national airspace system.

AI in the sky is bigger than SMART

SMART is only the tip of the iceberg. In September, Joby Aviation flew a zigzag route on a Cessna Caravan: from Concord, California, through Phoenix, Fort Worth, and South Carolina to North Carolina, about 3,200 miles. The aircraft navigated around thunderstorms and maneuvered through dense airspace on its own. A safety pilot was on board, but, according to the company, he did not need to intervene.

Money is driving all of this. Paying a pilot to fly a handful of passengers in an electric air taxi (eVTOL) is too expensive, so the business model only works if the aircraft fly themselves. Amazon plans to expand drone delivery to 500 sites across the US by the end of the year, and Walmart to 270 stores by the end of 2027. In the most ambitious scenario, market participants say, hundreds of thousands of drones will serve most of the population. Europe is moving more slowly: after a protest from the pilots' union, the European Union Aviation Safety Agency (EASA) postponed the idea of single-pilot airliner operations last year.

What this means in practice

For those building AI products in critical domains, the lesson is simple. A model that is right most of the time is not ready on that basis alone: you need an interface in which the operator quickly notices a strange recommendation and can reject it. It is also worth honestly accounting for the organizational effect: if deploying AI becomes a reason to cut the people who catch the model's errors, system reliability drops, even when the model itself is good.

Caveat: authorities have not found any link between SMART and the Friday the 13th outage, so there is no basis yet for blaming the system. The real question is different: will a human be able to catch its error when one does occur.

Story

Key Questions Answered

What SMART does and who is behind it?

The name stands for Strategic Management of Airspace, Routes and Trajectories. The program was developed by Boston-based Air Space Intelligence, and the FAA announced its deployment in September. It is predictive AI: it processes large volumes of flight data to warn in advance about scheduling problems and airspace congestion. FAA Administrator Bryan Bedford says that, together with equipment upgrades, SMART will help reduce delays, optimize traffic, and keep capacity open without compromising safety.

What are the key details on Two risks: model error and fewer people?

The first risk will be familiar to anyone who has worked with language models. If the program produces nonsense, will the operator notice? About two dozen Democrats in Congress reportedly wrote to the US Government Accountability Office asking it to review SMART. They fear the system could "hallucinate safety-critical information." The second risk concerns staffing. The US has a chronic shortage of controllers, and the FAA hopes AI will reduce the need for people.

What are the key details on AI in the sky is bigger than SMART?

SMART is only the tip of the iceberg. In September, Joby Aviation flew a zigzag route on a Cessna Caravan: from Concord, California, through Phoenix, Fort Worth, and South Carolina to North Carolina, about 3,200 miles. The aircraft navigated around thunderstorms and maneuvered through dense airspace on its own. A safety pilot was on board, but, according to the company, he did not need to intervene. Money is driving all of this.

ZK
Zhemal Khamidun — CPO at Alpina Digital, product manager of AlpinaGPT. Building AI products since 2019. Hamidun News is a pet project that grew into a production aggregator with 6,000+ readers.
✦ Unified AI Workspace · Nomad

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.

Try Nomad AI →Launch Nomad Agent
What do you think?
Loading comments…