AI pilots take flight as aviation tackles shortage and safety challenges

A small Cessna Caravan lifted off the runway with the pilot's hands nowhere near the controls. Tim Burns, watching from the back seat, radioed forward with a note of amusement: he wanted to see those hands stay free. The test pilot, Matt Diamond, sat to the left but was not flying the plane. Merlin Labs, an American startup, had handed the job to artificial intelligence.

I was, in legal terms, an experimental subject—the aircraft itself bore that label. The Merlin Pilot system does far more than a traditional autopilot. It uses natural language processing to listen to instructions from an air traffic controller and respond over the radio in a synthesized female voice. When Diamond said "Authorize," the plane banked toward a new heading, following orders it had understood and executed on its own.

For a pilot accustomed to holding the controls, surrendering them to a machine did not come naturally. But the demonstration mattered because aviation companies across the industry are turning to AI to reshape how planes fly—automating pilot tasks and, eventually, enabling fully autonomous flight. The timing is urgent. Boeing estimates that airlines worldwide will need more than 600,000 new pilots over the next two decades. Simultaneously, air traffic control systems are buckling under congestion, strained by a series of near-collisions and fatal accidents that have drawn public scrutiny and regulatory alarm.

The push toward AI-assisted aviation has found backing in Washington. Transportation Secretary Sean Duffy has championed artificial intelligence as part of the Trump administration's broader effort to modernize the nation's aging air traffic control infrastructure. "We will never outsource our national airspace to AI tools," Duffy told CNN. "Controllers will control the airspace, but we can make their job easier." The government sees AI as a way to lighten the load on controllers and improve efficiency in an increasingly crowded sky.

Merlin's argument is straightforward: eighty percent of aviation accidents still stem from human error. If the technology can reduce that figure, the company's CEO Matthew George said, it represents a meaningful use of engineering effort. The idea remains contentious. Commercial aviation has steadily incorporated automation for decades, culminating in fly-by-wire systems where computers interpret pilot commands even during manual flight. But Mykel Kochenderfer, whose research at Stanford University focuses on autonomous systems and aviation safety, notes a crucial difference. Modern AI-assisted systems are designed to handle a wider range of unexpected situations than traditional rule-based automation. "Our experience suggests this could be a very promising way to increase safety," he said, "but the industry still has a long road ahead to refine the technology and build the trust needed for acceptance."

Shifting pilot attitudes will not be simple. Current flight automation keeps the pilot at the center, allowing intervention when needed. Captain Jason Ambrosi, president of the Air Line Pilots Association representing more than 79,000 pilots in the United States and Canada, insists that automation and AI must support pilots, not replace them. "Technological advances can improve aviation safety, but they will never substitute for well-trained, rested pilots in the cockpit," Ambrosi said. Merlin pushes back against the specter of pilotless passenger flights. "We're not going to push a button and have crewless airplanes," George said. "This is about putting AI alongside human pilots and building trust."

The company has completed hundreds of test flights while pursuing Federal Aviation Administration certification—a process known for its rigor, often requiring years of testing and redundancy analysis before approval. The military may provide the first major proving ground. Merlin recently secured a contract worth more than $100 million with the U.S. Air Force to eventually bring the technology to C-130 cargo aircraft. As the Merlin system guided our plane toward the runway on final approach, it initiated a gradual descent and held the flight path steady despite a crosswind, touching down smoothly. "It's a challenging problem for automation," Diamond said as we taxied back to the hangar. "But once you solve it, things get much easier for the pilot."