Explaining AI Decisions to Air Traffic Controllers

Anduel Mehmeti, Gabriella Gigante, Salvatore Venticinque · July 24, 2026

A new study explores how to make AI decisions understandable for air traffic controllers, a high-stakes environment where trust is critical. Researchers trained a reinforcement learning agent to reroute flights around no-fly zones in a simplified air traffic control simulation. To explain the agent's choices, they used a saliency map, which highlights which input features (like distance to no-fly zone or fuel cost) most influenced each decision. This is a first step toward building transparent AI systems that can collaborate safely with human controllers. The practical implication: if AI can explain its reasoning, controllers can verify and trust its suggestions, potentially reducing workload and improving safety. Economic benefits include more efficient routing and lower fuel costs. However, the work is preliminary—real-world ATC is far more complex. Future research must test these methods in realistic, high-fidelity environments.