It’s one of the most casually repeated myths in aviation — that modern aircraft “fly themselves.” For some, it’s a comfort. For others, it triggers anxiety about losing human control. Either way, it’s based on a profound misunderstanding of how autopilot systems actually work. The belief that pilots are just “passengers in the cockpit” has no basis in reality — and it couldn’t be more wrong.
Autopilot is a critical system in aviation safety and efficiency, but it is not autonomous. It cannot make decisions, assess emergencies, or adapt to rapidly changing situations. Every action it performs is based on logic programmed by pilots, overseen by pilots, and if needed — interrupted by pilots. The truth is far more reassuring: autopilot is a tool, not a captain. Pilots are always in command.
This article explains in depth how autopilot functions, why it exists, what it actually does during flight, and why no commercial airliner is ever “flying itself.”
Autopilot: What It Is — and Isn’t
Autopilot, known technically as an Automatic Flight Control System (AFCS), is a computerised system designed to control the aircraft’s flight path based on commands from the flight management system (FMS) and direct inputs from the pilots. It can adjust the aircraft’s pitch, bank, altitude, speed, and heading. It can follow a pre-set route, climb to assigned altitudes, and even carry out precision approaches to a runway. But that’s where its intelligence ends.
Autopilot has no awareness of its environment. It cannot “see” weather ahead. It cannot understand nearby traffic. It cannot anticipate an engine failure, or react intuitively to a sudden wind shear. It operates purely on mathematical input. It will do exactly what it’s told — nothing more, nothing less. If that input is wrong, the autopilot will follow it blindly, which is why pilot oversight is not just essential — it’s constant.
Take-Off and Landing: Manual by Design
One of the most misunderstood aspects of automation is when it’s used. Take-off is always flown manually. Pilots handle the thrust settings, rotation angle, climb gradient, and aircraft configuration by hand. It’s only once the aircraft is safely airborne and stable — typically after the initial climb between 1,000 and 5,000 feet — that autopilot may be engaged.
Landing is also overwhelmingly manual. Unless conditions require an autoland (which we’ll address later), pilots disconnect the autopilot during descent or final approach. The last few thousand feet of altitude — especially the flare and touchdown — are flown by feel, judgement, and real-time inputs.
Even during cruise, automation is not left to run without supervision. Pilots are scanning instruments, monitoring weather, checking fuel calculations, tracking air traffic, and communicating with multiple air traffic control sectors. Autopilot may be flying the wings, but pilots are flying the mission.
Pilots Use — Not Rely On — Autopilot
The idea that automation replaces the pilot is backwards. In reality, automation frees the pilot to do more critical tasks. Rather than hand-fly a perfectly level cruise at 38,000 feet for hours, pilots use autopilot to maintain precise flight parameters while they manage weather diversions, coordinate with air traffic control, prepare for descent, or monitor systems for early signs of trouble.
Pilots also constantly cross-check what the automation is doing. They verify that altitude changes are correct. They monitor speed trends. They mentally rehearse each phase of flight so they’re prepared to step in if something deviates from normal.
The cockpit interface allows pilots to engage or disengage different autopilot modes at will. They can fly using heading mode (manual directional control), vertical speed mode (climb or descend at a set rate), LNAV (following a lateral route), or VNAV (managing vertical profiles based on waypoints and speeds). Each of these requires pilot input, and none of them makes the aircraft truly “autonomous.”
Pilots Are Required to Train Without Automation
Commercial pilots don’t just know how to fly without automation — they’re required to demonstrate it. Every six months, airline pilots are put through simulator sessions where key systems are disabled, including autopilot, autothrust, and flight directors.
They practise hand-flying from take-off to landing, often under abnormal or emergency conditions. These include engine failures, bird strikes, rapid decompression, electrical loss, and more — all while flying without the aid of computer systems. It’s not optional. It’s a regulatory requirement for licence retention and airline qualification.
This ensures that pilots never become “automation dependent.” If every system were to fail, they would still be able to fly the aircraft entirely by hand, using raw data and manual navigation techniques — just as pilots did for decades before automation was introduced.
Autoland Is the Exception — Not the Norm
It’s true that some aircraft can land automatically under specific conditions. This procedure, called autoland, is only used during extremely low-visibility landings at specially equipped airports. The aircraft must be certified for Category III operations, the runway must support it, and both pilots must be qualified to conduct it.
Even then, autoland is closely monitored. Pilots configure the aircraft, verify that approach signals are captured correctly, ensure the systems are functioning, and are fully prepared to go around if anything appears unstable. The aircraft will not taxi itself after touchdown. It will not stop on its own without pilot input. And it will not handle sudden gusts or runway contamination without pilot correction.
The vast majority of landings are done manually — because pilots are trained to do it better.
Failures, Deviations, and Disengagements
If anything goes wrong, autopilot doesn’t “decide” what to do. It disengages, or alerts the pilots, or continues executing an incorrect command until it’s overridden. This is why pilots are trained to instantly recognise automation errors. One incorrect mode selection, one wrong waypoint entry, or one sensor anomaly can lead to altitude or course deviations unless caught early.
There are countless examples of pilots intervening in automation failures. Whether it’s a weather radar malfunction causing a sudden course change, or a pitot tube icing up and feeding bad data into the autopilot, human judgement is the safeguard.
Even something as routine as turbulence may require autopilot disconnection. Pilots may hand-fly the aircraft through unstable air to better manage control inputs, vertical speed, and responsiveness — something autopilot can’t do intuitively.
Human + Machine: The Real Formula
The safest flights are not those where the aircraft “flies itself.” They’re the ones where human expertise and automated assistance work in partnership. Pilots use autopilot to reduce repetitive workload — but they retain decision-making at every step. They know the aircraft’s limits. They anticipate issues. They check weather, request alternate routes, adjust fuel strategies, and reroute to safer airspace as needed.
When automation works, it’s thanks to the crew’s understanding. And when it doesn’t, the crew are ready to fly — fully, manually, and decisively.
So next time you hear someone say, “Nobody’s flying the plane,” remember the truth: autopilot might be steering the aircraft’s controls for a moment, but trained professionals are flying it — with intention, precision, and constant awareness.
Disclaimer
For full legal, medical, psychological, and technical disclaimers relating to all content on this website, please refer to The Cockpit King’s official disclaimer page. All information is provided for educational and informational purposes only.