Rio de Janeiro, August 14, 2126.
Atlântica Flight 207 was scheduled to depart from Tom Jobim International Airport at 9:10 p.m., bound for Lisbon. The journey would take just under four hours.
Outside the terminal, the aircraft barely resembled the airplanes that had crossed the Atlantic a century earlier. It had no conventional vertical tail and no large engines hanging beneath its wings. Its fuselage blended seamlessly into the wings, creating a continuous lifting surface designed to produce maximum lift with minimum drag.
Ten small electric propulsors distributed along the trailing edge powered the aircraft. Their energy came from fuel cells supplied with liquid hydrogen stored in cryogenic tanks at the rear of the airframe. Structural batteries built into the wings provided additional power during takeoff and served as an emergency reserve.
Atlântica Flight 207 would carry 286 passengers, four cabin crew members and only one pilot.
Captain Lia Azevedo, 42, disliked it when people said she alone was responsible for the flight.
“We are never truly alone,” she would reply.
In addition to the human crew, the aircraft was equipped with an artificial intelligence system named Icarus. It was never officially described as a pilot, captain or copilot. Aviation authorities had prohibited artificial systems from receiving human titles.
Icarus was formally classified as a Level Three Certified Operational Assistance System.
In practice, it monitored every component of the aircraft, interpreted weather information, negotiated trajectories with the global airspace management network and assisted the captain with operational decisions.
But it had no legal authority over the flight.
That responsibility remained human.
Lia entered the flight deck and placed her hand on the instrument panel, just as she had done on her first day aboard the aircraft.
Traditional instruments had disappeared. There were no screens permanently divided into airspeed indicators, artificial horizons or navigation displays. The entire surface in front of the pilot could present information, but only the data relevant to each phase of flight appeared.
“Good evening, Captain,” Icarus said.
“Good evening. Aircraft status?”
“Airworthiness confirmed. Fuel cells operating at 98.7 percent efficiency. Energy reserves above regulatory minimums. No outstanding maintenance items.”
“Weather?”
A three-dimensional representation of the Atlantic appeared before Lia.
“Two areas of significant convective activity are forecast north of the planned route. An eastbound jet stream is also shifting south. I recommend an initial southerly trajectory followed by a progressive climb to flight level 610.”
Sixty-one thousand feet.
In 2026, that altitude would have been the domain of military aircraft, scientific balloons and a handful of experimental projects. In 2126, it was a common operating band for high-speed intercontinental flights.
“What is the projected saving?”
“Eleven minutes of flight time and a 3.2 percent reduction in hydrogen consumption.”
“Forecast uncertainty?”
“Approximately six percent during the first two hours. Less than three percent afterward.”
Lia studied the data for a few seconds.
“Route approved.”
Icarus transmitted the trajectory to the air traffic management network. No controller individually monitored every aircraft. Thousands of trajectories were coordinated continuously by automated systems under the supervision of human teams working in regional centers.
Each aircraft transmitted its position, speed, intended trajectory, endurance, performance and technical condition. Airspace was no longer organized solely around fixed airways. Temporary corridors were created and dissolved according to demand, weather and energy consumption.
Clearances no longer arrived by voice.
“Trajectory accepted,” Icarus announced. “Number three in the departure sequence. Estimated takeoff time: 9:12 p.m. and forty seconds.”
Atlântica Flight 207 left the gate without a tug. Small electric motors installed in the landing gear moved the aircraft almost silently.
At nearby terminals, electric regional aircraft were boarding passengers for cities that, a century earlier, had depended almost entirely on highways. Autonomous aircraft transported cargo. Urban air mobility vehicles connected the airport with different parts of the metropolitan region.
None interfered with the others. Every vehicle was part of the same network.
At exactly 9:12 p.m. and forty seconds, power increased automatically. The aircraft began its takeoff roll.
Lia kept her hands close to the controls, although she did not need to touch them.
The system adjusted the power of each propulsor independently. The wing surfaces subtly changed shape, adapting their aerodynamic profile to the aircraft’s speed and weight. There were no conventional flaps. The geometry of the wing itself changed during takeoff.
The aircraft lifted from the runway almost unnoticed by the neighborhoods surrounding the airport. Its noise footprint was far smaller than that of the jet engines used by previous generations.
“Landing gear retracted. Climb stabilized,” Icarus announced.
Lia watched Guanabara Bay disappear beneath a thin layer of clouds. The city became a collection of distant lights before fading beyond the horizon.
Forty minutes later, Atlântica Flight 207 was climbing through 50,000 feet.
Icarus controlled the aircraft, monitored energy consumption, updated the weather and compared the behavior of every system with records collected from millions of flight hours.
The flight deck remained silent.
To some passengers, flying had become as ordinary as boarding a train. They rarely thought about aerodynamics, navigation or operational safety. Many did not even know that a pilot was still on board.
Then an amber light appeared in front of Lia.
“Captain, I have detected a disagreement between the external atmospheric information and the aircraft’s optical sensors.”
“Nature of the disagreement?”
“Meteorological platforms report no significant particles. However, the onboard sensors are detecting an increasing concentration of particulate matter ahead.”
The three-dimensional display showed a yellowish area crossing their route.
“Volcanic ash?” Lia asked.
“Estimated probability: 41 percent. Atmospheric dust: 32 percent. Simultaneous sensor malfunction: 18 percent. Other causes: nine percent.”
“Possible source?”
“No significant eruption has been reported within the past six hours.”
Lia enlarged the image. The area did not appear to be following the forecast winds.
“What is the reliability of the external data?”
“Normal.”
“And the aircraft sensors?”
“Each sensor, considered independently, is reporting normal reliability. The disagreement cannot currently be explained.”
Within seconds, Icarus had analyzed satellites, oceanic platforms, nearby aircraft and meteorological models. Even so, it could not determine what lay ahead.
“Request data from aircraft that have crossed the area.”
“Already requested. Three aircraft recorded no particulate matter. One detected a slight reduction in propulsion efficiency, initially attributed to atmospheric variation.”
Lia remained silent.
Artificial intelligence could process more information than any human flight crew in history. Yet the volume of available data did not eliminate uncertainty. It merely allowed uncertainty to be observed with greater clarity.
“I recommend maintaining the present trajectory,” Icarus said. “The calculated probability of hazardous exposure is below 12 percent.”
“And if the external data are wrong?”
“There is insufficient evidence to support that conclusion.”
“But our sensors are seeing something.”
“Correct.”
Lia looked at the display.
That night, a solar storm had briefly interfered with several meteorological satellites. The global system had marked the information as valid because the received values remained within their expected limits. The data were not absurd.
They were simply old data repeated as though they were current.
Icarus did not know that yet.
But Lia noticed something different.
The indicated position of the particles was not changing according to the modeled winds because the wind information was also based on delayed data. The aircraft’s sensors, however, were observing the actual atmosphere.
“Icarus, temporarily disregard the regional satellite data. Recalculate using only local sensors, inertial navigation and reports from other aircraft.”
Two seconds later, the image changed completely.
“New analysis complete. Probability of volcanic ash: 87 percent.”
“Source?”
“Possible unreported submarine eruption approximately 900 kilometers northwest of the Azores. I recommend an immediate 23-degree right turn and descent to flight level 490.”
“Execute.”
The aircraft entered the turn smoothly.
“Notify air traffic management and transmit an alert to all nearby aircraft.”
“Message sent. The network is recalculating North Atlantic trajectories.”
Within minutes, dozens of flights changed course. Observation satellites were repositioned. An unmanned research aircraft was assigned to investigate the region.
The global system took less than four minutes to recognize the danger.
But the first decision had been human.
“Deviation established,” Icarus announced. “Captain, your decision contradicted my initial recommendation.”
“Yes.”
“What was the determining factor?”
Lia considered the question before answering.
“You tried to determine which source had the highest statistical probability of being correct. I tried to determine which error would produce the most serious consequence.”
Icarus remained silent for several seconds.
“Difference understood. The decision will be incorporated into the learning report following validation.”
“Do not incorporate it as a rule,” Lia replied. “Record it as a question.”
“What question?”
“What happens when the information everyone trusts is wrong?”
At 2:58 a.m. local time, Atlântica Flight 207 began its approach to Lisbon.
The airport no longer had a conventional control tower. Distributed sensors tracked every vehicle, person and moving object. Runway lighting adjusted according to the aircraft’s position, while autonomous inspection vehicles continuously examined the pavement.
The approach was flown automatically down to 200 feet.
“Visual conditions,” Icarus announced. “Runway confirmed. Automatic landing available.”
Lia watched the lights ahead.
“I have control.”
“Captain, the automatic landing system provides a higher precision margin than manual operation.”
“I know.”
“Is there an operational reason to take control?”
Lia smiled.
“There is a professional one.”
She placed her hands on the controls. The aircraft responded immediately. Despite all the automation, she could still feel the small variations in the wind, correct the alignment and control the flare.
The wheels touched the runway gently.
For a few seconds, Lia experienced the same sensation felt by the earliest aviators: the meeting between a machine, the air and the human determination to remain in the sky.
As the passengers disembarked, a child waited near the flight deck door. He was about nine years old and carried a small model aircraft in his hands.
“Are you the pilot?” he asked.
“I am.”
“But did you fly the airplane, or did the computer fly it?”
Lia looked back into the flight deck. The displays were dark. Icarus was silently organizing the technical records from the flight.
“We flew together.”
“Then is the computer a pilot too?”
“No. It can calculate, compare and predict far better than I can. But someone still has to decide what to do when the answers are not enough.”
The child looked down at the model aircraft.
“Will there still be pilots a hundred years from now?”
Lia smiled.
“As long as there is something unknown beyond the horizon, the sky will still need them.”
Marcuss Silva Reis
Commercial Fixed-Wing Pilot | General Aviation Pilot | Aviation Expert Witness | Economist | Optical Technician | Postgraduate qualifications in Aeronautical Sciences, Civil Aviation Safety and Higher Education Teaching | Former Civil Aviation Flight School Instructor | Former University Professor
Founder of Instituto do Ar
