FAA Prepares $875 Million AI System for Air Traffic Management

The US Federal Aviation Administration is preparing to deploy an artificial intelligence system that will advise on air traffic management in the Washington, DC, region. The planned launch would be the first limited step toward a potential nationwide rollout covering the airspace overseen by the FAA.

Key points

• SMART is intended to forecast traffic flows and identify possible conflicts using airline schedules, weather, airport capacity, and current airspace conditions.
• The initial deployment could cover the three major airports in the Washington area. The system is expected to provide alternative route information through existing FAA systems rather than change controller or airline procedures.
• SMART is included in a 12-year contract worth $875 million that the FAA awarded to Air Space Intelligence. The contract also includes a Flow Management Data and Services system intended to replace the current platform used at the FAA’s Air Traffic Control System Command Center.
• The FAA has not explained whether SMART relies primarily on deterministic models, statistical machine learning, or another form of AI.

The architecture appears to separate the operational foundation from the predictive layer. The new data and services platform would provide the backbone, while SMART would sit above it and estimate how traffic may develop. The FAA’s stated goal is to give the agency, airlines, and aircraft operators a common view of operating conditions. That shared picture could help them coordinate routes and departure times, with the broader aims of reducing fuel consumption, improving punctuality, and accelerating recovery after weather disruptions or congestion.

Why a narrow launch matters

Philip Mann, an aviation safety and AI governance consultant who previously worked at the FAA, argues that the central risk is not one bad prediction in isolation. It is the uncertainty involved in introducing a national-scale system with AI components that the agency has not previously operated. Limiting the first deployment reduces the number of unknowns and creates an opportunity to examine the system under real workload conditions.

Observers will be watching the boundaries of the trial. The FAA previously described a possible focus on traffic at 24,000 feet and above, including cruise flights and the climbs and descents feeding the Washington-area airports. The more important question may be what evidence is required before the system expands: performance under degraded data, interaction with actual operational workloads, and a written account of who owns the outcome when a recommendation is wrong.

The rollout comes while the FAA is already replacing aging radar, communications, and voice-switching equipment. The agency also faces a long-running shortage of air traffic controllers, with staffing and training pressures complicating its modernization efforts. An advisory AI system could help coordinate flows, but it is not a substitute for qualified controllers or resilient infrastructure.

Accountability will therefore be as important as prediction quality. A fatal 2025 collision involving an American Airlines jet and a US Army Black Hawk near Reagan Washington National Airport put renewed attention on staffing and airspace management in the region. For SMART to progress beyond a pilot, the FAA will need more than a successful demonstration: it will need credible safety validation, degraded-data testing, and clear responsibility when human operators act on imperfect advice.

Source: Ars Technica AI