This interview features Skyryse founder Mark Groden, who argues that general aviation's 'quiet crisis'—300-400 deaths per year—can be eliminated by SkyOS, an operating system using deterministic AI to make flying as safe as commercial aviation. Groden is optimistic, saying the technology is ready and adoption is the only hurdle. Key holdings: Skyryse (private, plans to deliver first SkyOS-equipped aircraft in 2025); Honeywell (a tier-1 supplier with high-margin spare parts, ~80% gross margin); Boeing/Airbus (commercial aviation duopoly, already very safe).
Skyryse founder and CEO Mark Groden is committed to raising general aviation safety to the level of commercial aviation through the universal flight system SkyOS, with the goal of reducing aviation fatalities to zero. Helicopter accident rates are significantly higher than those of fixed-wing aircra
Mark Groden is the founder and CEO of Skyryse, holding a PhD in Bayesian networks (deterministic AI). The main theme of this issue: general aviation faces a "quiet crisis" of 300–400 deaths per year, and Skyryse’s SkyOS operating system — an expert system that encodes industry expert knowledge into deterministic AI — has the potential to reduce that number to zero. The probability that all Americans aged 18 to 65 will die in a general aviation accident in their lifetime is more than two orders of magnitude higher than the probability of dying in a commercial aviation accident, and Groden believes this is not a law of physics, but a result of technological choices.
Groden emphasizes that the safety gap between general aviation and commercial aviation is enormous and has long been overlooked by the public.
Historical Context and Magnitude Comparison: Commercial aviation (Part 25) is the safest mode of transportation globally, second only to elevators. Meanwhile, general aviation (non-military, non-commercial aircraft and helicopters, about 400,000 units) experiences approximately 1,300 accidents annually in the United States, resulting in 300-400 deaths. Groden calls it the 'quiet crisis': 'Planes crash every week, and technology and high-level automation could have prevented them.'
Mechanism Breakdown: The safety of commercial aviation comes from a combination of 'highly skilled pilots + high automation level + large ground and air teams.' Its core is the digital fly-by-wire system: there is no mechanical connection between the pilot's control stick and the aircraft's control surfaces; instead, commands are transmitted via copper wire to a computer, which then controls the actuators. The computer is in the loop and provides safety envelope protection — for example, preventing the aircraft from stalling. When Sully Sullenberger landed on the Hudson River, it was this system that prevented him from over-pulling the stick and stalling, enabling a smooth landing.
Comparison: General aviation still uses mechanical linkages from the Wright brothers' era — the control stick is directly connected to the control surfaces via cables/push-pull rods, with no computer in the loop. Flying it is 'like driving a go-kart or riding a bicycle.' The control method for helicopters has not changed in over half a century since Igor Sikorsky's first flight.
Primary Cause of Accidents: The leading cause of fatal accidents in general aviation is loss of control in flight — the aircraft itself is intact, but the pilot experiences spatial disorientation, inputs incorrect commands, quickly exits the safe flight envelope, and enters an unrecoverable state. Although engine failure ranks high, aircraft can glide and helicopters can autorotate (using rotor inertia) to land safely.
Unique Data: Data from the U.S. helicopter safety team shows that after a helicopter enters a cloud, the average time from entering the cloud to a fatal accident is only 56 seconds. This explains the Kobe Bryant accident — an intact helicopter crashed into a mountain due to spatial disorientation.
Groden believes SkyOS is not a simple autonomous driving system, but "the first general aviation operating system ever," at its core a deterministic AI expert system.
Technical Architecture: SkyOS is a set of "huge equations that would fill an entire room if written out." It encodes the expert knowledge of the entire industry – knowing what every aircraft should do in every possible scenario, and "if we write out the equations and track all decision trees and defensive statements, we can show what happens next." Groden emphasizes this is a "responsible way to create technology": "We only pick up known and certifiable Lego bricks, then assemble them in a way that has never been done before, and write an operating system around it."
Human-Machine Interface Breakthrough: The team iterated through dozens or even hundreds of human-machine interaction methods. The final solution is one joystick + two touchscreens, and it is universal across all aircraft and helicopters. In contrast to the current state: every aircraft/helicopter cockpit is different, professional pilots typically hold qualifications for only two types, and muscle memory cannot be transferred. Groden's goal is to make it like a car – no matter which car you drive, the steering wheel, accelerator, and brake are all the same.
Validation Results: A person with zero flight experience, after 15 minutes of training on a simulator, can "fully and autonomously control the entire flight from takeoff to landing in a real full-size aircraft." In contrast, Groden recalls that when he first drove a car, "he couldn't even park, let alone drive in a parking lot."
Insights from Comparison with Cars: Cars became popular because all of us can drive them ourselves, without needing Max Verstappen (F1 driver) in the passenger seat. In the helicopter sector, out of about 20,000 helicopter pilots in the U.S., Groden estimates only about 1,000 are those he would be willing to ride with – equivalent to "if you want to drive, you must have an F1 driver with you." SkyOS's goal is to allow ordinary people to fly themselves, thereby fundamentally changing the unit economics: every seat is a paying passenger, and no longer requires a driver who does not participate in the trip.
Groden's assessment of the "flying car future" diverges sharply from the mainstream narrative: what can truly fly is not a car, but a helicopter transformed by new technology.
Viewpoint 1: A flying car is an engineering paradox. "Trying to build a flying car essentially means building a drivable airplane — the two engineering requirements are completely contradictory, and the result will be a terrible car and a terrible airplane, or both."
Viewpoint 2: Existing helicopter/airplane forms already have all the potential to realize the "flying car future," but are held back by safety. "Fundamentally, these platforms are already very efficient at converting stored energy into flight. We solved the physics problem, and we did it well. The issue is how to use today's technology to improve it."
Viewpoint 3: The first step on the path is to "build the market," and the prerequisite for building the market is solving safety. Groden supports this with a set of data: current general aviation annual production is about 2,500 units, with a historical peak (1970s) of about 17,000 units. In contrast, the Ford Model T was produced at a rate of 25,000 units per day in 1925. Groden asks new employees a trivia question: "No one has guessed more than 1,000." He concludes: "If Tesla or another automaker built cars at the current general aviation industry's production volume, those cars would be million-dollar items — as expensive as the prototypes."
Quantitative cost breakdown: Cars are cheap because of scale. After dissecting aircraft costs, Groden finds that a Ford Fiesta or Toyota Camry carries more onboard computing power and systems than "the most fully loaded general aviation airplane or helicopter" — yet cars are cheaper because they are built in large volumes. Once helicopters become safe, their idle time will drop: currently, about 25% of Los Angeles weather (cloudy conditions) prevents helicopters from taking off, while safety technology could make 99% of weather conditions flyable, fundamentally changing the unit economics.
Infrastructure perspective (counterintuitive): "Sky subway stations" are already built — there are thousands of general aviation airports across the U.S., most built 50 years ago, and the majority of people live within a 20-minute drive of a general aviation airport. Moreover, "there are no delays on the sky track" — flight paths are GPS waypoints with no maintenance costs. Airspace throughput is "almost unlimited because you have three dimensions."
Groden believes that automation is not about replacing the pilot, but about freeing the pilot from "constantly balancing marbles on an inverted bowl with both hands and feet," allowing them to focus on higher-level decision-making.
Current dilemma: A hovering helicopter is like "trying to place a small marble on top of an inverted bowl, and then the pilot uses two index fingers to keep poking it to maintain balance." The pilot has one hand on the cyclic, one hand on the collective, and both feet on the pedals — all occupied, with extremely high cognitive load. Any distraction could lead to disaster.
Future vision: The best pilots already act like "mission commanders" — they think about mission objectives and how to execute them, rather than getting bogged down in the details of specific system management. "It simply doesn't make sense to use humans to troubleshoot all systems." SkyOS's goal is to give pilots "the level of a 10,000-hour pilot," whether they are novices or experts.
Impact on pilot economics (counterintuitive): Groden argues that pilot income will rise, not fall. Reasons: (1) Weather availability increases from 75% to 99%, meaning a significant jump in aircraft utilization and a broader income source for pilots; (2) Cross-airframe versatility allows pilots to serve more missions; (3) More people flying → more aircraft → more demand for pilots — "a rising tide lifts all boats."
Long-term outlook (10-20 years): "In the future, teenagers can get their driver's license and pilot's license at the same time. In terms of complexity, technology has already made flying easier than driving." Groden believes this "will take decades to achieve," but the current generation of children "look at our cockpits and operating systems, and they think it's already a reality" — while the actual cockpit "looks like something from World War II."
Groden analyzes the economic logic of aircraft manufacturers (OEMs), revealing why SkyOS's business model can achieve market adoption.
OEM's true profit source: OEMs have extremely low profit margins on selling new aircraft—single-digit percentages. The real profit comes from servicing the existing installed base: life-cycle services such as spare parts, overhaul, and maintenance, typically priced at 3x the original unit, with gross margins of ~80%, and the OEM is locked in due to airworthiness certification. Therefore, "OEMs want the largest fleet and the most flight hours." Every accident hurts everyone—undermines public confidence, and the blame falls on the aircraft. "No one is willing to risk their lives on a mode of transportation."
Market validation model: Groden points out that historically, safety-enhancing technologies have achieved 100% market adoption in aviation very quickly. The reason is that "people are willing to pay a premium for a safer vehicle."
Tier 1 supplier leverage: Tier 1 suppliers such as Honeywell have significant bargaining power over Part 25 airliner OEMs—if that box doesn't work, the aircraft cannot take off. Honeywell has life-cycle contracts, with spare parts priced at 3x the original, and is locked in by certification. But general aviation OEMs "are not good at manufacturing actuators, automation systems, and even less skilled at writing software and operating systems"—this is exactly the entry point for SkyOS.
Groden provides a clear timeline for the current stage: the technology is complete, the product is complete – a state reached only in the past 12 months. The goal is for people to 'get the keys to a SkyOS-equipped aircraft' next year (2025).
Evolving self-awareness: In the early days of the company, the team thought they needed 'fully autonomous flight' and indeed achieved the first unmanned helicopter flight (no safety pilot) three months after inception. But Groden quickly realized, 'Building an unmanned aircraft is not difficult. The difficulty is elevating that technology to an extremely high safety level and capturing human intent and skill in real time.' He stated clearly, 'There is no truly "fully autonomous" vehicle in use today – it is always about the combination of human intent and machine automation.'
Falsification condition: If market adoption lags behind expectations, Groden believes that is the only risk. But the motivation is 'we can truly influence industry safety,' and he believes the pilot community (existing and aspiring pilots) will adopt it 'fairly quickly.'
| Target | Guest View | Key Data |
|---|---|---|
| Skyryse (unlisted) | Under construction, founder fully committed | SkyOS has been in development for 7-8 years; target key delivery in 2025; unmanned helicopter first flight 3 months after company founding |
| Honeywell | Background mention (Tier 1 supplier business model) | Spare parts priced at 3x original, gross margin about 80% |
| Boeing / Airbus | Background mention (Part 25 airliner duopoly) | Commercial aviation is the safest mode of transportation per seat mile |
| Embraer | Third largest in commercial aviation | Not discussed in detail |
| F-35 Program | Mentioned as an analogy | $500 billion development project, achieved unified cockpit concept |
1. “After a helicopter enters a cloud, a fatal accident occurs on average in 56 seconds” — Mark Groden cites data from the U.S. Helicopter Safety Team to explain the root cause of the Kobe Bryant accident: spatial disorientation.
2. “Building an unmanned aircraft is not difficult; the difficulty lies in elevating the technology to an extremely high safety level and capturing human intent and skill in real time” — Groden cites the fact that the company achieved its first unmanned helicopter flight three months after its founding, but notes that the real 7–8 years of effort were spent on achieving “human+machine” synergy, rather than full autonomy.
3. “We currently have no truly ‘fully autonomous’ vehicle in use at any scale” — Groden defines autonomous driving as “the combination of human intent and machine automation” and points out that autonomous cars are “the hardest application scenario,” because “a single block in New York City is more complex than a cross-continental flight.”
4. “A person with zero flight experience, after 15 minutes of simulator training, can take full control of a real helicopter from takeoff to landing” — This is direct evidence of the effectiveness of the SkyOS human-machine interface design, and empirical proof of Groden’s claim that “flying has already become easier than driving.”
5. “Of the approximately 20,000 helicopter pilots in the U.S., only about 1,000 are ones I would be willing to fly with” — This is equivalent to “to drive a car, you must have F1 driver Max Verstappen in the passenger seat,” revealing the fundamental constraint that current pilot skill scarcity imposes on the market.
6. “The Ford Model T was produced at a rate of 25,000 units per day in 1925, while the general aviation industry today produces 2,500 units per year” — Groden uses this figure to illustrate that aviation has never achieved economies of scale, which is also the root cause of high helicopter prices. (Note: Model T daily production is 3,650 times the current annual production of general aviation — on an annualized basis.)
7. “OEM margins are in the single digits when selling new aircraft; the real profit comes from spare parts and services — priced at 3x the original parts, with 80% gross margins” — This explains why OEMs are fully incentivized to adopt safety technology to expand fleet size, rather than resist it.
8. “The general aviation cockpit looks like it’s from World War II; but when children see our operating system, they think this is already reality” — Groden uses a generational perception gap to illustrate that the technology itself is no longer the bottleneck; the mental model is.
9. “The best pilots are already ‘mission commanders’; they think about mission objectives, not system details — using humans to troubleshoot all systems simply doesn’t make sense” — Groden’s core philosophy: automation is not about replacing humans, but about letting humans do what humans do best.
10. “Airspace throughput is almost unlimited because you have three dimensions; and ‘aerial subway stations’ (general aviation airports) are already built, with most people living within a 20-minute drive” — Countering the common objection of “inadequate infrastructure,” pointing out that the real bottleneck is safety, not physical space.