Introduction

Very recently, a friend of mine who is Captain on an Airbus A320 told me half-jokingly but half-seriously too that “We are not flying planes but managing them”. This assertion is ever so true in light of the exponential increase of AI technology in general and Agentic AI in particular. Later this year, for example, the Institute of Aerospace Medicine will be hosting a symposium in Hamburg, Germany to present the final stages of the so-called LOKI project which is a collaborative effort to advance trustworthy human-AI collaboration in aviation. Topics that shall be discussed include: Digital Interactive Reliable Controllers, the Intelligent Pilot Advisory System, Control Center Task Environment, Applied Eye Tracking measurements and psycho-physiological equipment.

The power of AI in aviation

The integration of artificial intelligence (AI) into aviation is often misunderstood as a pathway toward fully autonomous aircraft that replace human pilots. In reality, the advancement of technological development of AI suggests a different future: one in which pilots’ role remain central, but their role evolves into supervising increasingly sophisticated automated systems. The influence of AI on pilot behaviour and performance is therefore less about replacement or substitution and more about transformation, particularly in how pilots interact with information and data in making decisions and managing workload. It is no state secret that modern aircraft already rely heavily on automation. Systems such as autopilot, flight management systems (FMS), and flight envelope protections have been standard for decades. The benefits are huge. For instance, such systems can monitor thousands of parameters at once, detect abnormal trends earlier than humans, predict failures before they happen and suggest or prime actions during abnormal situations. From a pure psychological perspective, humans cannot process so much complex data at once and even if they were, they would potentially ‘select’; a process we refer to as ‘satisficying’; meaning that when there is overwhelming pieces of information and data, we arbitrarily select which information to retain and focus on, assuming it will give us best returns to our actions. The choice is not rational and very often quite random. The bottom-line is therefore that such systems are less likely to allow human error and increase the chances to make aviation even more safe as a means of transport. Within this scenario though (at least for the foreseeable future), the pilot will remain legally and operationally in command.

These technologies allow aircraft to maintain course, altitude, and efficiency with minimal manual input. As a result, pilots today spend less time physically flying the aircraft and more time managing and monitoring (scanning) systems. This shift has already altered pilot behaviour, emphasizing cognitive skills such as system oversight, situational awareness, and decision-making rather than manual control alone.

The introduction of AI therefore builds upon this foundation by enhancing decision support rather than assuming control. AI systems are designed to process vast amounts of data in real time, far beyond human capability. They can, unlike us, monitor thousands of parameters simultaneously, detect subtle abnormalities, and identify patterns that may indicate future system failures. For pilots, this means earlier warnings and more informed insights during both routine operations and abnormal situations. Consequently, pilot performance is likely to improve in terms of accuracy, response time, and risk management.

Behavioural challenges

However, this increased reliance on AI also introduces new behavioural challenges. One key concern is automation dependency, often termed as high reliance or over-reliance. As systems become more capable, pilots may be tempted to rely too heavily on them, potentially reducing their manual flying skills and their ability to respond effectively in the event of system failure. This requires a shift in pilot attention and awareness: Maintaining a balance between trust in AI and active engagement with AI remains critical. Training programs will need to adapt, ensuring that pilots retain core flying competencies while also developing the skills required to interpret and question AI-generated recommendations.

In the near future, AI is expected to further reduce pilot workload, especially during high-stress scenarios such as diversions or emergencies. AI-assisted fault diagnostics can help identify the root cause of technical issues more quickly, while real-time optimization tools can suggest more efficient routes based on weather and traffic conditions. These capabilities can free up cognitive resources, allowing pilots to focus on strategic decision-making rather than being overwhelmed by information processing.

Despite these advancements, however, the pilot remains both legally and operationally in command of the aircraft. This principle is fundamental to aviation safety. Unlike other industries, aviation does not adopt new technologies solely because they are advanced or intelligent. Instead, systems are only trusted when their behaviour, including their potential failure modes, is fully understood and predictable. This cautious approach limits the extent to which AI can independently control aircraft, particularly in passenger aviation.

There is, however, potential for reduced-crew operations in cargo aviation, where regulatory and safety considerations differ slightly due to the fact that no passengers are involved (except the crew). Even in these cases, though, implementation would require extensive redundancy, oversight, and approval from aviation authorities. Fully pilotless passenger flights are not currently part of any approved roadmap, a state of affairs that reflects both technical limitations and public trust considerations.

Conclusion

Ultimately, the future cockpit will not be empty, but redefined. It will be a space where human expertise and AI work in partnership not in competition. Pilots will act less as manual operators and more as system managers, decision-makers, and safety overseers. This evolution will demand new skills, including data interpretation, system understanding, and critical thinking, while still preserving the fundamental human judgment that underpins aviation safety. In conclusion, AI’s impact on pilot behaviour and performance is profound but not disruptive in the sense of replacement. Instead, it enhances capabilities, reshapes responsibilities, and introduces new challenges. The enduring presence of the human pilot ensures that aviation remains not only technologically advanced but also fundamentally human-led.