Will AI Replace Pilots? A Personal Story from the Cockpit Dreamer Generation
Now at fifty, when I look back on my life, two professions stand out as roads not taken: airline pilot and firefighter. Firefighting strikes me as a genuinely noble calling—a career where meaningful public service, a deep sense of pride, and a reliable livelihood all converge in a single vocation. As for flying: on evening jogs through a darkening park, I sometimes catch the distant throb of a propeller somewhere above me. The aircraft itself is invisible against the dark sky, yet that faint sound has always filled me with a reverence I find hard to explain.
The moment that crystallized my admiration for aviation came during my years of military service in my mid-to-late twenties. We were resting after an individual combat drill one clear November morning at the training center when an F-16 fighter jet passed directly overhead, letting out an ear-splitting roar. The cockpit, visible as the aircraft rolled against the blue sky, struck me as the purest expression of engineered beauty I had ever seen. It was less the raw thrill of speed than the aircraft's sculptural elegance that gripped me. From that day on, I followed aviation films, magazines, documentaries, and airshows with a devotion that has never faded.
In the mid-twentieth century, the flight deck of a large long-haul airliner could carry as many as five specialists—a captain, a first officer, a flight engineer, a navigator, and a radio operator—sharing responsibility for a single journey. Today, on wide-body commercial jets, that crew has been reduced to two: a captain and a first officer. Advances in computer automation gradually absorbed the roles of the others. Now, with artificial intelligence accelerating at an unprecedented pace, we may be entering an era in which even that two-person minimum is no longer considered technically necessary.
AI vs. Human Pilots: The First Dogfight Between an AI Jet and a Human-Flown F-16
The transition toward AI-controlled flight has already moved well beyond simulation labs and into live combat testing. According to reporting by DefenseScoop and The Aviationist, the U.S. Air Force's X-62A VISTA program—developed in partnership with DARPA—carried out what the Department of Defense described as the first within-visual-range dogfight ever conducted between an AI-controlled jet and a human-piloted F-16. Across 21 test flights flown between December 2022 and September 2023, AI agents flew the X-62A in nose-to-nose engagements against a human pilot—a milestone the department publicly disclosed in 2024. Safety pilots remained on board throughout, with the authority to intervene at any point.
During these engagements, the two aircraft closed to within approximately 2,000 feet of each other at speeds exceeding roughly 1,200 miles per hour—conditions that approach the outer limits of human reaction time.
The Aviationist noted that the X-62A functions as a variable-stability flying testbed. The program advanced from installing live AI agents on the aircraft to a live dogfight demonstration within a single year—a development pace that would have seemed implausible only five years earlier. These flights did not eliminate human oversight from the architecture; they showed how capable AI had become within it. I keep coming back to that closing distance—2,000 feet, at well over a thousand miles an hour. Having once stood near enough to a fighter to feel the air change in my chest, I find it hard to picture two of them threading that gap on software's command without equal parts awe and unease.
A Chilean Air Force F-16A Block 20 MLU fighter jet banks in flight.
How AI Handles Emergency Landings When Pilots Are Incapacitated
For commercial aviation, automated systems are being engineered primarily to manage the scenarios human crews dread most: the sudden incapacitation of the pilot in command. Airbus's DragonFly project—a research and demonstration initiative, not yet a certified commercial system—uses computer vision and an array of onboard sensors to detect pilot incapacitation and respond autonomously. According to Airbus's official program documentation and coverage by The Register, the system can execute route diversions, manage ground taxi procedures, communicate with air traffic control—media reporting has described this as including synthetic-voice transmissions—and perform a safe landing at the nearest suitable airport, all without a conscious pilot at the controls.
DARPA's Aircrew Labor In-Cockpit Automation System (ALIAS) takes a different approach: a removable, drop-in retrofit kit designed to reduce pilot workload across a wide range of existing platforms. According to DARPA program documentation and reporting by Interesting Engineering, ALIAS has been tested aboard a modified Sikorsky S-76B helicopter and in a simulated Boeing 737 environment—not an operational passenger aircraft—with the supervising pilot monitoring the system from a tablet rather than the controls.
Critically, neither program is designed to eliminate the pilot from the flight deck. Both are positioned as fail-safe companions—systems engineered to step in precisely when human performance degrades during the highest-stakes phases of flight. As a passenger, I find that idea both reassuring and faintly unsettling: comforted that an aircraft could bring itself down safely if the worst happened, uneasy at how seldom we are told such systems exist at all.
AI Predictive Maintenance: Catching Failures Before They Happen
Before an aircraft ever requires emergency intervention, machine learning is already working quietly in the background, analyzing mechanical health in real time. The Airbus Skywise platform ingests the vast streams of sensor data modern aircraft produce—on the order of a terabyte of data per day for the latest-generation jets—and cross-references it against historical maintenance records to predict component failures before they occur in service. According to Airbus, nearly 12,000 aircraft are now connected to the Skywise network—a figure that reflects how deeply data-driven maintenance has taken hold across the commercial aviation industry.
Airlines using Skywise apply its natural language processing capabilities to flag repetitive fault patterns across entire fleets. Operators have proactively avoided technical delays by identifying imminent failures in integrated drive generators, high-pressure valves, and radio altimeters before those components ever reach a failure state in flight.
Aerial refueling of a USAF B-2 Spirit stealth bomber by a RAAF KC-30A Multi-Role Tanker Transport.
Why AI Cannot Fully Replace Pilots: Regulatory Limits and Fatal Blind Spots
Despite these technological achievements, regulatory bodies have drawn firm lines around how much authority automation may hold over a commercial flight. After a multi-year study of reduced-crew concepts, the European Union Aviation Safety Agency (EASA) concluded that, with current cockpit designs, an equivalent level of safety to today's two-pilot operations cannot be sufficiently demonstrated—and paused any move toward single-pilot airline service. Its evaluations underscored that a lone pilot cannot reliably manage the full flight envelope during abnormal or emergency events—the exact scenarios where a second trained mind matters most.
| Concept | EASA Definition |
|---|---|
| Extended Minimum Crew Operations (eMCO) | Permits single-pilot operations during cruise phase only, with the second pilot resting in a designated crew rest area. |
| Single-Pilot Operations (SiPO) | Enables end-to-end single-pilot operations for the entire flight, from pushback to gate arrival. |
The consequences of granting automated systems covert authority over flight controls are not hypothetical—they have already been measured in lives. The Indonesian National Transportation Safety Committee (KNKT) and the U.S. National Transportation Safety Board (NTSB) both documented how the Boeing 737 MAX's Maneuvering Characteristics Augmentation System (MCAS) relied on input from a single angle-of-attack sensor and would repeatedly force the aircraft's nose downward without the crew's command or full understanding of what was occurring. When I read through the findings, what stayed with me was not the technology but the silence around it—crews fighting a system no one had told them was there.
Boeing significantly expanded MCAS's authority over the flight controls during the program's development, and in March 2016 it requested—and the FAA approved—removing all reference to the system from the Flight Crew Operations Manual. The KNKT accident report on the Lion Air crash, and the NTSB's joint investigation findings, establish that active 737 MAX pilots received no training on the system's existence before the Lion Air accident. Federal certification guidelines assumed a pilot would counter such a runaway-stabilizer condition within roughly four seconds—yet Boeing's own simulator testing had shown a test pilot taking more than ten seconds to diagnose and respond, a delay the company's analysis flagged as potentially catastrophic. The system did not fail because it was autonomous—it failed because its authority was hidden.
Whether AI across all industries ultimately serves humanity or threatens it is a question that history, extensive public debate, and further technological development will answer together. The honest answer is that both possibilities are real and neither should be dismissed.
What is clear is this: we cannot hand every phase of flight to machines and walk away. The Boeing 737 MAX disaster did not prove that pilots can always overcome a flawed automated system. It proved something more precise and more sobering—that when an automated system acts outside the pilot's awareness and overrides human inputs without transparency, the result is catastrophic. Even if AI one day manages 99.9% of flight efficiency, the pilot's role in confronting the unpredictable 0.1%—and in retaining final authority over the aircraft—remains a domain that no algorithm should inherit by default.
In the Star Wars films, an X-wing fighter carries both a pilot and an astromech droid like R2-D2. Theirs is a relationship of strict symbiosis: neither replaces the other. Having grown up watching those films and spending decades following real-world aviation with equal devotion, I find myself hoping our future looks something like that. Today, I want to express my genuine admiration for the pilots who bear responsibility for their passengers and their nations with every flight they take.
For more, visit www.thesecom.net.
Frequently Asked Questions
What is the X-62A VISTA, and what did it prove about AI pilots?
The X-62A VISTA is a heavily modified F-16 operated by the U.S. Air Force Test Pilot School as a variable-stability flying testbed. Developed in partnership with DARPA, the program ran 21 test flights between December 2022 and September 2023, culminating in the first within-visual-range dogfight ever staged between an AI-flown jet and a human-piloted F-16—closing to within approximately 2,000 feet at speeds exceeding roughly 1,200 mph. The Department of Defense publicly disclosed the achievement in 2024. Safety pilots onboard retained the ability to intervene at any moment. The program showed that AI can operate within the combat flight envelope—while also demonstrating that human oversight remains a deliberate part of the system architecture. (Sources: DARPA, U.S. Air Force Test Pilot School, DefenseScoop, The Aviationist.)
How does Airbus DragonFly assist pilots?
The Airbus DragonFly system is a research and demonstration project—not a certified commercial product—that uses sensors and computer vision to detect pilot incapacitation. Once triggered, it can autonomously execute route diversions, coordinate with air traffic control via synthetic voice, manage ground taxi, and perform an emergency landing at the nearest suitable airport, all without a conscious pilot at the controls. (Sources: Airbus official program documentation; The Register.)
Why did EASA stop short of approving single-pilot airline operations?
Following a multi-year research study (its eMCO-SiPO project), the European Union Aviation Safety Agency concluded that, with current cockpit designs, an equivalent level of safety to today's two-pilot operations cannot be sufficiently demonstrated—and it paused progress toward reduced-crew commercial operations. The agency pointed to unresolved challenges around pilot-incapacitation monitoring, fatigue, and emergency response, recommending that new "smart cockpit" technologies be developed and tested first. (Source: EASA eMCO-SiPO Safety Risk Assessment, final report 2025.)
Sources & References
- DARPA – X-62A VISTA & ALIAS program documentation
- U.S. Air Force Test Pilot School – X-62A program reports (2024)
- DefenseScoop – AI dogfight coverage (2024)
- The Aviationist – X-62A flight test reporting (2024)
- Airbus – DragonFly & Skywise official program pages
- The Register – Airbus DragonFly coverage
- Interesting Engineering – DARPA ALIAS reporting
- European Union Aviation Safety Agency (EASA) – eMCO-SiPO Safety Risk Assessment Framework (final report, 2025)
- Indonesian National Transportation Safety Committee (KNKT) – Lion Air Flight 610 Final Accident Report
- U.S. National Transportation Safety Board (NTSB) – Boeing 737 MAX investigation findings
- U.S. House Committee on Transportation & Infrastructure – Boeing 737 MAX certification investigation (MCAS authority, FCOM removal, crew response time)
- ICAO / FAA – Historical flight crew complement documentation
📷 Image Sources & Credits
- Photo 1 (FACh F-16A Tail #747): F-16.net Global Aircraft Database / Scramble Dutch Aviation Society. Chilean Air Force (Fuerza Aérea de Chile) F-16A Block 20 MLU '747' (ex-RNLAF J-864) / Peace Amstel I Program.
- Photo 2 (RAAF KC-30A & USAF B-2 Spirit): U.S. government public domain image via Wikimedia Commons. "A39-005 refuelling a B-2 Spirit in August 2022" — a RAAF KC-30A Multi-Role Tanker Transport refueling a USAF B-2 Spirit, August 2022.
Comments
Post a Comment