AI Won’t Replace Operators in Aviation. Here’s What It Will Actually Do.
I have run operations with sophisticated AI models. The best decisions came when experienced planners saw the patterns AI surfaced — then applied context the model couldn’t know.
- The displacement narrative around AI in aviation misunderstands what expertise actually consists of in complex operational environments.
- At DHL, the best decisions came when an experienced planner saw a pattern the model surfaced, then applied context the model couldn't know.
- The 52% eBooking adoption at Polar came from giving human judgment better tools, not from replacing it.
The dominant narrative around AI in aviation and logistics is one of displacement: AI will optimise routing, predict demand, replace the network planners and commercial teams who currently make these decisions. I have been running operations with sophisticated AI and predictive models for the better part of a decade, and the displacement narrative is wrong — not because AI lacks capability, but because it misunderstands what expertise actually consists of in complex operational environments.
At DHL, we had predictive models running across 41 countries. The models were good — they identified patterns in demand data, flagged capacity imbalances, surfaced pricing anomalies. But the best decisions we made were not the ones where the model told us what to do. They were the ones where an experienced network planner saw the pattern the model surfaced, then applied context the model couldn't know: a government policy change opening a market in 48 hours, a competitor's operational disruption creating a two-week capacity window, a customer relationship that made a below-model yield acceptable because of long-term strategic value. These are not data problems. They are judgment problems.
“The model surfaces the pattern. The operator understands the context. Together, they make decisions that neither makes well alone.”
The 52% eBooking adoption we achieved at Polar did not come from replacing human judgment with automated systems — it came from giving human judgment better tools: real-time visibility, better demand data, faster feedback loops. The humans still made the decisions; they made them faster and with more confidence. What AI actually does in complex operational environments is expand the decision surface available to operators. It surfaces patterns that would take a human analyst days to identify, removes the cognitive load of monitoring routine parameters, and lets experienced operators focus their attention where judgment matters most. This is the same augmentation logic behind the Convergent Technologies framework. The organisations that will get the most from AI investment are not the ones deploying it as a replacement for operational expertise. They are the ones using it to augment operators who already know what good judgment looks like — and giving those operators a larger surface to apply it.