Skynet won’t solve aviation’s problems, humans will
Mark Groden is the founder and CEO of Skyryse, the developers of SkyOS, a Deterministic Expert AI-based universal operating system for flight.
December 11, 2024
Read MoreI founded Skyryse, the developer of SkyOS, a Deterministic Expert AI-based universal operating system for flight, over eight years ago. Our first aircraft with SkyOS, the Skyryse One, is the world’s first highly-automated fly-by-wire helicopter with a single control stick.
I’ve spent my entire life working on making an automated future a reality and what I’ve found is that there is no way to safely adopt non-deterministic, user-replacing fully autonomous systems at scale in aviation.
No matter how you look at it, human oversight is essential to achieving the safest outcomes in flight. From setting mission objectives to assessing real-time risks, human involvement in decision-making is critical.
"The US military should only put service members in unpiloted aircraft when parents are comfortable putting their children in driverless cars on the way to soccer practice."
- Lt. Gen. (Ret) Ross Coffman
The good news is there’s no need to remove the human from the loop in passenger-carrying flight for a simple reason. The human is already on the aircraft. So instead of removing agency from the human onboard, let’s take advantage of that.
Self-driving cars require removing the human driver because nearly everyone can drive, at little cost to learn, and therefore the only economic benefit comes from removing that driver. In aviation, piloting skills are limited to a small percentage of the population, and the cost to learn is extremely high. Despite the high cost and significant training, every accident that has ever occurred still involves one of this small percentage of people who have spent the time and money to become a highly-trained pilot.
But, with highly automated systems – like SkyOS – we can make piloting more accessible, providing pilots with a higher level of safety and greater control of the aircraft with simplified, aircraft-agnostic controls. It lets oversight evolve to pilots making high-level decisions rather than controlling minute flight mechanics that require hours and hours of mind-numbing training.
The additional benefit is that it cuts training costs, giving pilots more time to increase the number of aircraft they’re certified to fly – allowing more people to pilot aircraft in most every environment, democratizing aviation with practical solutions.
Given that, if we can unlock the perceived economic benefits of autonomy by enabling anyone who can drive to "pilot" safely, even without 10,000 hours of developed expertise, you can get similar cost benefits without having to remove the human from the loop.
That doesn’t mean we always need to have a human onboard. For certain geofenced missions without passengers or with a direct human risk (e.g., firefighting), we’d still want to keep a human pilot involved in the operation of the aircraft even if they’re not sitting in the pilot’s seat. But, by placing human oversight offboard through remote operation we can offer safer, more effective operations, allowing the aircraft to take on greater risk where necessary and removing the risk of the human onboard.
In these situations, we’d still keep the human as a decision maker for a simple reason – there’s no AI or autonomous system that can yet replace a human’s ability to make the right call in evolving environments. That intuition, adaptability, and experience is key in aviation, particularly during unpredictable moments or evolving objectives.
Even in these unmanned situations, the systems that fly the aircraft must be engineered to the most stringent levels in aviation today. Otherwise a failure onboard can result in a crash outside the geofenced area and the range of possible negative consequences could extend to the range of the aircraft.
That’s why we’ve designed our SkyOS system with a balanced approach – high levels of automation to assist humans with routine, error-prone tasks behind the controls, and remote operation for high-risk, unmanned missions. The result will be that anyone can pilot any aircraft.
This is the only way to successfully, simply, and safely scale flight for any mission.