Quonset State Airport, Rhode Island —
As a small Cessna Caravan picks up speed on the runway and ascends into the sky, the pilot next to me maintains a relaxed demeanor, hands entirely off the controls.
“Show us those jazz hands,” Tim Burns, the chief technology officer of Merlin Labs, quips over the airplane's intercom from the rear seats.
On this particular flight, Matt Diamond, the test pilot, is not at all in control of the aircraft. A sophisticated artificial intelligence system now manages the essential functions of flying, surpassing the capabilities of a standard autopilot. Utilizing advanced natural language processing, the Merlin Pilot system listens to commands from a simulated air traffic controller and responds through a digital voice. When test pilot Diamond states, “Authorize,” the plane adjusts its heading as if on cue.
As someone who flies myself—and who tends to be somewhat controlling—relinquishing control to a machine was an unfamiliar experience. However, the demonstration holds significant implications as the aviation industry increasingly turns to AI to revolutionize air travel, aiming to automate various pilot tasks and potentially facilitate fully automated flights in the future.
Test pilot Matt Diamond (right) lets the AI take over as the aircraft accelerates for takeoff, with CNN's Pete Muntean observing.
This test flight occurs during a time when airlines globally are grappling with a significant shortage of pilots. Boeing projects that the aviation sector will require over 600,000 new pilots in the next twenty years. Concurrently, air traffic control systems face mounting stress amidst growing air traffic and a series of alarming close calls and accidents.
The increasing integration of AI in aviation is being supported by governmental figures as well. Transportation Secretary Sean Duffy has advocated for AI tools as part of the broader initiative to modernize the nation’s aging air traffic control framework.
“We will never completely hand over the national airspace to AI,” Duffy stated in a recent CNN interview. “Air traffic controllers will remain in charge, but we can certainly ease their workload.”
Duffy indicated that AI could be leveraged to alleviate pressures on controllers while enhancing efficiency in an ever-busier airspace.
Merlin Labs believes that AI could similarly address cockpit challenges. “Human error accounts for 80% of aviation accidents,” remarked Merlin CEO Matthew George in a CNN interview. “If we can mitigate that, we have a valuable opportunity to make a difference.”
However, this vision is not free of contention. Commercial aviation has gradually embraced increased automation over the years, evolving into today's fly-by-wire systems where computers translate pilot inputs, even during manual operations.
“The current automation in modern cockpits serves a limited range of functions,” explained Mykel Kochenderfer, a Stanford University researcher focused on autonomous systems and aviation safety. He noted that emerging AI-driven systems are designed to manage a wider array of unpredictable scenarios compared to conventional rule-based systems.
“Experience indicates this approach may significantly enhance safety,” he asserted, “yet there remains a long journey ahead to refine the technology and cultivate the necessary trust for widespread acceptance.”
Changing the perceptions of pilots could be a formidable challenge. Present automation systems place pilots at the core of operations, enabling them to step in as needed.
Capt. Jason Ambrosi, president of the Air Line Pilots Association, representing over 79,000 North American pilots, emphasized that automation and AI should complement pilots rather than replace them.
“While technological innovations can bolster aviation safety, they will never substitute for the competent pilots aboard every aircraft,” Ambrosi stated to CNN. “Ultimately, the most crucial safety asset on any flight will always be two well-trained pilots in the cockpit.”
Merlin Labs highlights that fully autonomous passenger flights remain a distant goal. “We aren’t suddenly launching uncrewed planes,” George clarified. “This effort is about integrating AI with human pilots and establishing mutual trust.”
The company has completed hundreds of test flights as it seeks certification from the Federal Aviation Administration, an agency known for its strict safety standards that often require extensive testing and thorough analysis before allowing new systems into service.
The military could serve as a critical testing ground for the technology's first major applications. Merlin recently secured a contract exceeding $100 million with the U.S. Air Force, aimed at adapting the system for use in C-130 cargo aircraft.
As the automated system aligns us for landing, it begins a smooth descent toward runway 34, expertly managing the controls to maintain course despite a gentle crosswind right up until touchdown.
“It presents a challenging scenario for the automation,” test pilot Diamond remarks while we taxi back to Merlin’s hangar. “But once the solution is found, it certainly eases the pilot's workload.”
