A memory that’s stayed with me from my stint in the military nearly 50 years ago is a sign that my supply sergeant kept on his desk: “We’ve been working with less and less for so long that now we can do everything with nothing!” (His complaint: We never had enough replacement aircraft parts to satisfy the squadron’s needs.) I was a young Marine Corps aviator at the time, soon to leave the service for a pilot job with Pan Am—and in hindsight, looking back at the changes that have taken place during my time in commercial aviation, that sign now seems prescient, as some in the industry are beginning to ponder the idea of pilotless flights.

Today, twin-engine airliners like Boeing 777s and Airbus’s A330s can fly farther than earlier four-engine 747s with comparable passenger loads, while consuming roughly 40 percent less fuel—and they do it with two pilots instead of the three-cockpit crew of earlier airliners. The role of the traditional third crew member, the flight engineer, was eliminated in the 1980s, when controls were automated and placed in the pilot’s overhead panel. Today, extra pilots are only used on long international flights so that everyone on the crew can take some time to rest.

The latest airliner models are controlled largely through computers with automatically updated flight-management systems, commonly referred to as fly-by-wire systems. All this automation has eliminated many of the routine pilot functions of the past: setting courses, switching navigation radio frequencies, tending to miscellaneous other details that kept past pilots involved over the course of the flight.

These newer, more efficient models have much lower operating costs, but the downside is that pilots, increasingly removed from the mechanics of flying, no longer need to be as attentive. On a 2.5 hour domestic flight, autopilots and flight-management systems typically do about 95 percent of the work. For maximum efficiency, autopilots are typically engaged after takeoff, at about one or two thousand feet, and pilots don’t take over again until the plane is lined up on final approach, a few thousand feet above the airport.  

Of course pilots can manually fly airliners if they wish—and some do activate the autopilot later in the climb, or take over earlier during descent—but the norm is more use of automation, not less. It’s the way modern airliners are designed and intended to be flown.   

Computer-automation equipment that warns pilots of potentially dangerous flight conditions has made air travel vastly safer than when I was a young 707 copilot in the late 1960s, and statistics are getting better all the time. Last year, the International Air Transport Association reported no passenger jetliner accident fatalities anywhere in the world, excluding criminal acts.      

But no fatalities doesn’t mean no mistakes, and despite the industry’s best efforts to prevent human error, there have been incidents where crews ignored timely warnings, mishandled flight controls and system malfunctions or failed to notice oversights of other crew members. In October 2009, for example, pilots on a Northwest Airlines flight from San Diego to Minneapolis/St. Paul began an extended conversation unrelated to the flight after reaching cruise altitude. The distracted crew missed calls from air-traffic control and other flights in their vicinity, and finally failed to notice that they’d overshot the destination airport. They eventually reestablished radio contact and landed uneventfully, but the incident highlighted how high levels of automation can degrade attention and affect human performance.

Similar incidents have fueled government concern about the erosion of basic hand-flying skills and overall flight-crew awareness. In January 2016, the Office of the Inspector General at the U.S. Department of Transportation expressed concern to the Federal Aviation Administration, recommending that airlines improve their pilot-training and performance-monitoring protocols and develop standards for manual-flying skills. The FAA, which has been grappling for years with the question of how to keep pilots attentive when automated systems do just about all the flying,  promised to provide formal guidance by early next year.

Meanwhile, NASA is currently studying the concept of single-pilot airline cockpits, where a first officer on the ground is monitoring several flights at once. In emergencies or unusual circumstances, the ground first officer could transfer responsibilities for other flights and focus solely on the flight that needed attention; if the sole pilot becomes incapacitated, the plane could be controlled from the ground.

But follow this idea to its logical conclusion, and you reach the question: Why have a pilot in the air at all? NASA’s single-pilot concept sounds like it’d be prone to many of the same performance issues of highly automated two-pilot airliners, with the added twist of the pilot’s extended isolation for hours on end, locked behind the cockpit door.

In a talk at the American Institute of Aeronautics and Astronautics SciTech Conference in January 2015, John Tracy, Boeing’s chief technology officer, said: “Some of our freighter customers are asking us for those [autonomous airplane] systems today.”

He went on: “We are quite confident that technologically, the toolkit is filled. With respect to a commercial airplane, there is no doubt in our minds that we can solve the problem of autonomous flight. It’s a question of certification procedures, regulatory requirements and, even more significantly, public perception. Will the flying public be comfortable getting onto a commercial plane with no pilot?”

For the near future, at least, it seems that the question may remain a hypothetical. When asked if the company is developing an autonomous commercial plane, a Boeing spokesperson responded:  “Whatever research Boeing may be doing in this area is proprietary and we won’t elaborate beyond what Mr. Tracy said.”

A spokesperson for Airbus, Boeing’s fiercest competitor, said: “Airbus is not developing an autonomous airplane.”

And a spokesperson for the FAA, which would be responsible for certifying an autonomous plane, said: “The FAA has no current unmanned aircraft certification projects in the transport category, nor has anyone engaged the agency on such a project.”

But as Tracy put it, the toolkit is filled—the latest fly-by-wire commercial jets, like Boeing’s 787 and the Airbus A350, already have flight-control computers, so there’s no need to start from scratch. The design work is largely done. In emergencies, instead of pilots following pop-up instructions and monitoring systems on display screens, all steps could be accomplished automatically, as the autopilot flawlessly flew the plane on the appropriate route, altitude, and airspeed. Nice, at least in theory.

But when it comes to flying planes, humans are still better than computers at quickly assimilating unrelated facts and acting on them. Consider, for example, Captain Chelsey Sullenberger’s landing on January 15, 2009, when he successfully avoided a crash by navigating an Airbus A320 into the Hudson River. Or the United Airlines crew at Sioux City, Iowa, on July 19, 1989, artfully steering their crippled DC-10 to a survivable crash landing after the center engine exploded, disabling the plane’s redundant flight controls. When the first autonomous plane rolls out of the factory with no cockpit windows, I hope it can match the performance of those crews. Until that’s a possibility, the idea of “doing everything with nothing” will remain just an idea, and the cockpits will stay filled.