I bring you a weekly bite-sized chunk of the science behind helicopter human factors and CRM in practice, simplifying the complex and distilling a helicopter related study into a summary of less than 500 words.
TITLE:
Human–AI teaming in the cockpit: From theory to practice: Views from the flightdeck.
WHAT?
Study examining whether established human teamwork principles and Crew Resource Management (CRM) competencies remain applicable when artificial intelligence becomes an increasingly capable participant in cockpit operations.
WHERE?
Netherlands Aerospace Centre (NLR).
WHEN?
Published 29 September 2026 in Aerospace.
WHY?
CRM was developed around human–human teamwork. As AI systems progress from providing information towards actively supporting decisions and coordinating tasks, established concepts such as mutual trust, shared mental models and leadership may no longer translate directly.
The researchers wanted to establish whether existing frameworks could identify emerging human–AI teamwork risks and inform future cockpit design and CRM training.
HOW?
Researchers conducted two complementary qualitative expert reviews of five proposed cockpit AI applications: ATC speech-to-text, runway/taxiway prediction, diversion advice, emergency diagnosis and startle-response assistance.
Four human-factors experts assessed these applications using Salas’s Big Five Teamwork model, while four professional pilots with CRM expertise examined their implications for established pilot competencies.
Both groups considered potential benefits, vulnerabilities and whether existing teamwork frameworks required adaptation. The study evaluated AI use cases rather than directly observing pilot–AI teamwork.
FINDINGS:
Both teamwork and competency frameworks helped identify potential benefits and vulnerabilities associated with cockpit AI.
- In terms of adaptability and workload management, AI could improve crews’ ability to manage unfamiliar situations, generate alternatives and redistribute cognitive demands.
- xEffective collaboration requires pilots to understand what information AI is using, its recommendations and its limitations to develop shared understanding.
- Human–AI trust differs from interpersonal trust. Both excessive reliance and unnecessary rejection of AI advice could compromise performance.
- The concept of CRM would need development to incorporate AI dynamics. The experts identified potential training requirements involving critical evaluation of AI recommendations, resolving pilot–AI disagreements, communication and workload management.
- The authors also proposed refinements to teamwork theory, particularly around cognitive offloading, calibrated reliance and explainability.
SO WHAT?
The study suggests that AI will not simply introduce another cockpit tool. Instead, increasingly capable systems may change how crews coordinate, make decisions and distribute work.
This raises fundamental questions about the nature of CRM with AI systems. How should pilots challenge AI recommendations? How can they recognise when their understanding differs from the system’s? And how should workload and authority be managed when AI begins to take initiative?
Rather than replacing established non-technical competencies, the findings suggest extending their application to human–AI interaction.
However, this was an exploratory study based on preliminary expert judgements, not demonstrated operational outcomes. Further empirical research is needed to establish whether the proposed adaptations improve safety and performance.
REFERENCE:
Kox, E., van Rooij, J., Zon, R., & Kirwan, B. (2026). Human–AI teaming in the cockpit: From theory to practice—Views from the flightdeck. Aerospace, 13(10), Article 878. https://doi.org/10.3390/aerospace13100878
