The exercises aimed to assess how advanced AI support tools could assist both Air Traffic Controllers (ATCOs) and Supervisors in managing Flight-Centric and Single Controller Operations (SCO).

Over four days, we ran a series of structured and measured simulations to evaluate:

• The AI-driven allocation model (assigning and reassigning flights dynamically)
• The AI’s ability to detect and manage proximity and conflict situations
• Human-AI teaming in real time – both from the ATCO and Supervisor perspectives
• Situational awareness, workload, and trust through structured questionnaires and open debriefs
• How transparent and explainable AI decision-making processes really are – and what level is needed to ensure user confidence

Methodology:
Realistic measured exercises, operational scenario-based runs, live ATCO feedback, and human performance assessments (Bedford, SACHA, SATI), followed by extensive debriefs and tailored questionnaires.

This validation is a critical step to the path toward AI-augmented ATM systems, but also a reminder of how much more we need to question, test, and iterate. More work lies ahead to refine the operational roles, the AI-human interaction model, and the training pathways.

A big thank you to the controllers, engineers, researchers, and project leads who contributed to this collaborative work with their sharp insights, technical expertise, and critical observations. Looking forward to sharing more results soon.

To learn more about the project: hypersolver.recherche.enac.fr