AI and aviation: why human performance still matters most
Dr. Lea Trampitsch-Vink works at the intersection of human performance, aviation psychology and operational design, with roles at Austro Control, CANSO and the European Commission.
Her session at our ATSEP seminar last year stayed with us because it focused on how AI can support human strengths as complexity grows, and what that requires from the systems built around it. The session focused on how AI can support human strengths as complexity grows, and what that requires from the systems built around it.
She kept returning to the same point. Aviation still depends on human judgement, adaptation and communication, especially when situations become less predictable. That matters even more in a system under pressure from growing traffic, staffing shortages and new airspace users.
She illustrated that through a human performance pyramid. At the base sat culture. Above that came organisation, then selection and training, and finally the daily operational layer of workload, situational awareness, teamwork and supervision. The logic was clear. Human performance does not begin at the console. It is shaped much earlier, by the system around the person.
Her comparison was simple and effective: human performance is like fuel. It is consumed while we work. If tasks are unclear, tools do not fit, or complexity rises too far, that fuel runs down faster and the risk of error rises with it.
She used the CSAR model to show where AI can take pressure off the people doing the work. It can help gather information and identify patterns, while decisions and actions remain with the people responsible for them.
Her Synapses model looks at how complexity builds over time and how that affects workload and fatigue. The purpose is to spot rising pressure early enough to adjust breaks, workload or sector configuration before performance is affected.
This also challenged the usual focus on traffic volume. A count of aircraft per hour says how busy the airspace is, but not how demanding the work is for the people managing it. A slide contrasted aircraft per hour with cumulative complexity, arguing that traffic volume alone does not fully describe operational demand.
For an ATSEP audience, the relevance went beyond the controller role. Lea pointed to supervisors and flow managers as good starting points for AI-supported performance management, and linked the same thinking to ATSEP work, where predictive models could strengthen monitoring and system oversight.
The takeaway from the room was not that AI changes everything. It was that aviation needs a more serious understanding of human performance if it wants to use AI well. Measure the right things. Reduce cognitive overload where possible. Build support around the strengths people already bring.
The central point of the session was not less focus on humans, but a clearer understanding of what supports strong human performance.
Would you like to discuss how we can support your ATSEP training? Contact us.