FAA’s AI Push in Aviation Signals a Faster, Riskier Shift to Autonomous Airspace

The Federal Aviation Administration rolled out an AI-powered airspace tool called SMART in September, a move that underscores how quickly aviation is moving toward drone operations, eVTOLs and more automated traffic management. The new system is designed to help controllers predict scheduling and traffic problems before they ripple through the system, not to replace human oversight.

What the FAA has actually deployed

According to WIRED, SMART is an internal airspace tool meant to spot likely delays and congestion earlier, giving the FAA and its controllers another layer of forecasting at a time when the agency is under pressure from rising traffic complexity and mounting interest in autonomy.

The rollout came just after Transportation Secretary Sean Duffy publicly praised the tool on Sept. 27, calling it a

“GAME CHANGER.”

The FAA has framed SMART as a support system rather than a substitute for controllers, but the timing of the deployment has pushed it into a broader debate about whether AI should be trusted inside one of the most safety-sensitive parts of U.S. infrastructure.

How SMART fits into aviation’s bigger automation push

SMART is arriving as drone delivery firms and eVTOL developers push harder for routine access to the national airspace. That matters because aviation is not a sector that absorbs change quickly. Certification is slow, the consequences of failure are severe, and even small software or procedure errors can cascade into major disruptions.

The FAA is not starting from zero. The agency began waiving some rules in 2023 to allow beyond-visual-line-of-sight drone flights, a major step toward making drone delivery practical at scale. At the same time, the industry is trying to prove that autonomous aircraft can operate safely alongside conventional traffic.

Why it matters

This is not just about one new software tool. SMART sits at the center of a larger question: whether the FAA can use AI to improve safety and efficiency without creating new blind spots in the system.

For passengers, the concern is straightforward. If AI is feeding decisions inside the airspace system, regulators need to know whether it is reducing delays and workload or introducing fresh risk through bad predictions, overconfidence, or weak oversight. For airlines and controllers, the issue is whether automation helps ease staffing pressure or masks it.

The stakes rose further after a Friday the 13th groundstop triggered public anxiety about whether AI had anything to do with a communications failure. Administration officials said there was no evidence of a cyberattack, but Homeland Security secretary Markwayne Mullin would not completely rule out an AI failure, saying,

“There’s nothing certain right now.”

That uncertainty matters because public trust in aviation depends on clear answers when systems fail.

There is also a labor dimension. WIRED reported that the FAA cut its active controller staffing target in May from 14,633 to 12,563. Against that backdrop, software that promises better prediction and flow management may look less like an experiment and more like a necessity.

Joby’s long flight shows how quickly the industry is moving

Joby Aviation has offered one of the clearest examples of how far the sector is trying to move in practical terms. The company completed a 3,200-mile pilotless Cessna Caravan trip across the United States with a safety pilot aboard. The route ran from California to North Carolina, with stops or legs through Arizona, Texas and South Carolina.

Joby says the safety pilot never had to intervene. That is a powerful claim, even if it does not mean the flight was fully autonomous in the regulatory sense. It shows the industry’s preferred narrative: that automation can take on more of the work while human supervision remains available in the background.

The company is also using the FAA’s eIPP pilot program for demonstration flights in New York, California and Texas. In New York, Joby says it completed the first-ever point-to-point eVTOL flight within New York City, flying from JFK to Manhattan heliports. Those flights are important not because they are commercial service, but because they test whether new aircraft types can fit into crowded urban airspace without breaking the system around them.

Background: how aviation got here

The shift now underway did not begin with SMART. It grew out of decades of planning for a more digital air traffic system and a slower, often frustrating effort to move aviation beyond legacy constraints.

In the early 2000s, drones became more capable and officials began thinking seriously about shared airspace. The Next Generation Air Transportation System, launched in 2007, was supposed to modernize how the United States manages flights. ADS-B, the broadcast surveillance system aircraft use to share position and track data, was supposed to be part of that broader overhaul by 2020.

But aviation modernization rarely arrives in a single sweep. The industry has had to work around a set of long-running obstacles: certification delays, safety culture, public concern and the technical challenge of mixing autonomous or remotely piloted vehicles with conventional aircraft. That is why progress has come in pieces — drone waivers, demo flights, traffic-management software, and now AI tools for controller support.

Other regions have moved cautiously as well. WIRED noted that the European Union Aviation Safety Agency shelved a proposal from Airbus and Dassault for a single-pilot airliner after pilot union pushback. That episode is a reminder that even where technology exists, labor resistance and safety concerns can slow the pace of adoption.

What SMART is, and what it is not

SMART is not a sci-fi autopilot for the national airspace. It is a predictive system designed to help controllers anticipate bottlenecks, scheduling trouble and traffic snarls before they happen.

That distinction matters. Air traffic control is built on redundancy, judgment and procedural discipline. Any AI system dropped into that environment has to support decision-making without undermining accountability. If a tool suggests a course of action, the human controller still has to understand why it is being suggested and whether the machine is wrong.

That is why the reporting around SMART has drawn scrutiny from lawmakers. Democratic lawmakers reportedly sought a Government Accountability Office review of the system on Oct. 13 over hallucination concerns. Even if SMART is not a generative chatbot, the concern is the same: if a predictive system is wrong, who catches the mistake before it affects flights?

Who is asking the hard questions?

Lawmakers, regulators and safety experts are all implicated, but the most immediate pressure is on the FAA itself. The agency has to prove that the tool improves operations without becoming a hidden dependency. It also has to show that AI oversight can fit inside aviation’s deeply conservative safety framework.

Michael McCormick, a professor of air traffic management at Embry-Riddle Aeronautical University, captured the scale of the shift when he said

“The objective is not to use traditional air traffic control”

in the same way forever. The quote points to a bigger truth: the goal is not merely digitizing old habits, but redesigning how traffic is managed as the number and type of aircraft in the sky change.

By the numbers

Several of the clearest markers in this story are numerical. They show both the pace of change and the scale of the challenge.

Why airlines, regulators and passengers should care

For airlines, the practical question is whether more predictive tools can reduce delays, reroutes and ground disruptions. The FAA’s system is not just about future aircraft; it is about the daily management of congestion and the squeeze between rising demand and limited human staffing.

For regulators, the issue is more delicate. The FAA must balance innovation with the public expectation that air travel remains one of the safest forms of transportation. That means testing AI tools in a way that does not outpace oversight. If SMART becomes deeply embedded before it is thoroughly understood, the agency could find itself defending a system it cannot fully explain.

For passengers, the story may eventually become visible in small ways first. That could mean fewer delays, smoother traffic flow or faster approvals for new drone services. But it could also mean more debate whenever a system outage, communications failure or scheduling glitch happens and people wonder whether automation was part of the problem.

What happens next

The next phase is about scrutiny, not announcement. Democratic lawmakers’ push for a GAO review means SMART could face formal review over whether its AI functions are reliable and appropriately governed. The FAA will also have to keep showing that the tool is helping rather than creating operational confusion.

At the same time, the industry’s broader test cases will keep multiplying. Amazon plans to expand drone delivery to 500 U.S. locations by the end of the year, while Walmart is targeting drone delivery from 270 store locations by the end of 2027. Those ambitions depend on more than aircraft hardware. They depend on airspace rules, controller tools and the public’s tolerance for increasingly automated flight.

Joby’s next steps matter too. The company’s New York, California and Texas demo flights are evidence that eVTOL development is no longer confined to renderings and promises. The open question is when those demonstrations turn into routine commercial service, and how the FAA will segregate autonomous traffic from human-piloted aircraft while keeping the system safe.

There is no single deadline that resolves these questions. Instead, the aviation sector is moving through a series of controlled experiments, regulatory waivers and public tests. SMART is one of them. Joby’s flights are another. Drone delivery expansion is a third. Together, they point to an airspace that is becoming more automated before the public has fully absorbed what that means.

The bottom line

The FAA’s September deployment of SMART shows that AI is entering aviation first as infrastructure, not as a passenger-facing robot or a pilotless commercial jet. That may make the transition quieter than the science-fiction version, but it does not make it less consequential.

The central question now is whether the agency can use predictive AI to improve safety and efficiency without undermining the trust, transparency and redundancy that aviation depends on. The next round of reviews, demonstrations and rule changes will determine whether SMART becomes a foundation for the future of flight or another reminder that automation in aviation moves fastest when the risks are hardest to see.