FAA Invests in AI for Air Traffic Management

AI to Solve the Chaos in the American Skies

The US Federal Aviation Administration (FAA) is preparing a massive investment in air traffic control. The goal of an up to $875 million contract for the development of AI tools is to deploy the SMART system, which aims to optimize flight paths. The tool will first be deployed in the Washington D.C. metropolitan area, one of the most complex airspaces in the country, before eventually rolling out nationwide.

A Shared Picture Instead of Isolated Decisions

The key change the SMART project is supposed to bring is the unification of data. Artificial intelligence should generate forecasts and recommendations that the FAA, airlines, and aircraft operators will see simultaneously. Instead of each party making their own estimates of delays and weather, they will all share a single operational picture, allowing for faster agreement on the most efficient routes and departure times.

The Tool Won't Replace Missing Personnel

The SMART system is primarily intended to streamline fuel use and speed up the recovery of traffic from weather events. However, a crucial context is the critical and chronic shortage of air traffic controllers in the US. While the AI tool might help manage traffic with fewer people by optimizing workloads, it will not replace the real-time human decisions that maintain safe separations between aircraft.

The First Severe Winter Storm Will Be the True Test

The real benchmark for the hundreds-of-millions-of-dollars project won't be how smoothly it operates on a sunny day in Washington. The acid test will be how quickly it can recover the network after a major winter storm or a systemic outage. If the algorithm can't resolve cascading delays better than the current veterans in the control towers, it will amount to little more than very expensive visualization software.

Lilith's verdict

The problem with American aviation isn't a lack of data for dispatchers. The problem is they are short three thousand controllers. Even a billion-dollar software package won't put more people in front of radar screens.