The Real Causes of AI-171: Beyond Pilot Error

AI-171 was not a pilot error but a failure to train the unimaginable. In aviation, we do not rise to the occasion; we fall to the level of our training. Loss of Control In-Flight (LOC-I) remains one of the most catastrophic accidents, with 38 incidents between 2011 and 2014 alone claiming 1,242 lives, as reported by IATA.

These events, like the 2014 AirAsia crash, often occur in seconds, leaving no room for conscious decision-making — only habit, reflex, and training. While the industry has invested nearly a billion dollars in Upset Prevention and Recovery Training (UPRT), the uncomfortable truth is that the training duration, retention, and reinforcement are grossly inadequate to instil automatic, counterintuitive responses like pushing the stick forward during a stall.

Neuroscience, psychology, and decades of behavioural research confirm that forming such reflexes takes time, repetition, and reinforcement — far more than current programs allow. Attributing the AI-171 crash to pilot error is unjust and dangerously simplistic. It ignores the deep-rooted cognitive science of habit formation and the systemic shortfalls in training pilots for the rarest — and deadliest — scenarios.

The Loss of Control-In-flight Accident report by IATA analyses the 38 LOC-I accidents that occurred during 2011-14 and caused 1,242 fatalities and the contributing factors. LOC-I accidents like the 2014 Air Asia often have catastrophic results with very few, if any, survivors. Reducing the number of LOC-I accidents is an ICAO priority, and ICAO has developed harmonised training requirements for flight crews that address and mitigate LOC-I events.

The industry has spent a whopping 1,000 million USD on training pilots and trainers. Does this ensure that LOC-I accidents will be eliminated, as they were with the introduction of the Ground Proximity Warning System?

Training

Upset recovery training for the prevention of LOC-I is collated into an integrated approach that identifies the training resources—academic, on aeroplane, and Flight simulator-based. It also includes the associated elements of training required to provide pilots with the necessary knowledge, skills, and attitudes (KSA) to reduce the probability of an upset encounter.

This knowledge maximises their ability to recover from such an event.

Stall protection

The training aims to accurately and early recognise all available aural, visual, and tactile alerts to an approaching stall and, with due consideration to maintaining adequate safety margins, an aerodynamic stall. Particular attention is given to aeroplane stall characteristics without a stall warning indication.

Background information on the stall

The lift of an airfoil depends on its aerodynamic coefficient (Cl) and the square of the airflow speed. The aerodynamic coefficient increases with the angle of attack (noted as alpha) up to a maximum value, after which it decreases when the angle of attack continues to increase. This tipping point, where the aerodynamic coefficient is at maximum, is the marker, from an aerodynamic point of view, for the stall. The angle of attack at which the Cz is at a maximum is thus the stall angle of attack (alphamax).

The aerodynamic characteristics of an aerofoil, thus the evolution of the Cl = f (alpha) curve, are different between the lower layers (low Mach, subsonic airflow, incompressible air) and the high altitudes (higher Mach, airflow close to trans-sonic, influence of the air’s compressibility).

Understanding this can help in the design of more effective training programmes.

Recovery from an impending stall or after a stall requires the pilot to unload the wing or reduce the angle of attack. If, for example, the aircraft’s nose drops after a stall, the pilot would instinctively pull the control column back to raise the nose because this is what they do daily. In this scenario, the pilot will push the stick to fly the nose down and reduce the angle of attack.

How long does it take to form a new habit?

Stall recovery is a counterintuitive manoeuvre. Skill is learnt through repetition till it forms a habit and becomes automatic. Maxwell Maltz was a plastic surgeon in the 1950s when he noticed a strange pattern among his patients. When Dr. Maltz would operate — like a nose job, for example — he found that it would take the patient about 21 days to get used to seeing their new face.

Similarly, when a patient had an arm or a leg amputated, Maltz noticed that the patient would sense a phantom limb for about 21 days before adjusting to the new situation. These experiences prompted Maltz to think about his adjustment period to changes and new behaviours, and he noticed that it also took him about 21 days to form a new habit.

Maltz wrote about these experiences and said, “These, and many other commonly observed phenomena tend to show that it requires a minimum of about 21 days for an old mental image to dissolve and a new one to jell.”

Phillippa Lally is a health psychology researcher at University College London. In a study published in the European Journal of Social Psychology, Lally and her research team determined how long it takes to form a habit. On average, it takes more than two months before a new behaviour becomes automatic — 66 days to be exact. And how long it takes a new habit to form can vary widely depending on the behaviour, the person, and the circumstances. In Lally’s study, people took 18 to 254 days to form a new habit.

The current duration of training is grossly inadequate for the pilot’s behaviour to become automatic when approaching a stall/stall recovery.

Ebbinghaus forgetting curve

Ebbinghaus’ forgetting curve describes the decrease in the ability of the brain to retain memory over time. The issue was hypothesised by Hermann Ebbinghaus in 1885, which is why it’s called the Ebbinghaus forgetting curve. The theory is that humans start losing the memory of learned knowledge over time, in days or weeks, unless the learned knowledge is consciously reviewed repeatedly. A related concept to the forgetting curve is the strength of memory, which states that the period up to which a person can recall any memory is based on the strength of the particular memory.

Ebbinghaus found the forgetting curve to be exponential. When learning any particular piece of information, memory retention is 100%. However, it drops rapidly to 40% within the first few days, after which the decline slows again.

Furthermore, addressing the emotional and cognitive aspects of training is crucial for long-term retention and effectiveness.

The forgetting curve is exponential because memory loss is rapid and huge within the first few days of learning. However, the rate of memory loss decreases, and the rate of forgetting becomes much slower from then on.

Ebbinghaus also discovered another phenomenon called overlearning during his study of the forgetting curve. The basic idea is that if you practice something more than what is usually required to memorise it, the effect of overlearning takes place. This means that the information is now stored much more strongly, and thus, the effects of the forgetting curve for overlearned information are shallower.

Conclusion

A billion dollars spent on training should not vanish into previously never properly built muscle memory. When instinct and training collide, instinct often wins, unless we retrain the very core of that instinct.

We must stop treating UPRT as a technical module and start treating it as an emotional, cognitive, and profoundly human challenge that can’t be solved with a single simulator session.

The cost of inadequate training isn’t measured in currency. It’s measured in lives.

Only when training becomes transformation will we truly reduce the risk of Loss of Control In-Flight.

Upset Prevention and Recovery Training (UPRT) is not about checkboxes and compliance. It’s about rewiring reactions, creating new defaults under stress, and ensuring that the next pilot faced with the unimaginable doesn’t freeze, or worse, do the wrong thing, reflexively.

The cost of inadequate training isn’t measured in currency. It’s measured in lives.

We must stop treating UPRT as a technical module and start treating it as an emotional, cognitive, and profoundly human challenge that can’t be solved with a single simulator session.

Only when training becomes transformation will we truly reduce the risk of Loss of Control In-Flight.