The press release read like a script from a low-budget sci-fi thriller. The United States Air Force just accepted its first uncrewed fighter aircraft, planting a flag in the ground and declaring that the pilotless dogfight is finally here. Defense media rushed to print, parroting the Pentagon’s talking points about artificial intelligence dogfighting, loyal wingmen, and the twilight of the human cockpit.
It is a comforting narrative. It keeps congressional checkbooks open, gives contractors another decade of cost-plus funding, and allows aviation nostalgics to pretend that the twenty-first century air war is just World War Two with better Wi-Fi. If you found value in this post, you might want to check out: this related article.
It is also complete nonsense.
I have spent years watching defense procurement programs burn capital on shiny objects while ignoring the brutal physics of actual combat. Calling a remote or autonomous jet a fighter because it has hardpoints and a sleek silhouette is like calling a cruise missile a bomber. A fighter is not a shape; it is a cognitive anchor in a chaotic environment. Strip away the biological pilot who can feel G-forces in their teeth and make split-second ethical decisions under acute terror, and you do not have a fighter. You have a smart munitions bus with an identity crisis. For another perspective on this story, refer to the latest coverage from Engadget.
The lazy consensus in military technology right now is that autonomy is simply a drop-in replacement for human biology. Swap the carbon-based pilot for a silicon chip, keep the exact same tactical doctrines, and watch the aerial dominance roll in. This is a fatal category error.
The Physics Of Attrition Versus The Myth Of The Dogfight
Let us look at the fundamental metrics that military planners love to obscure behind buzzwords. A traditional crewed fighter like the F-35 is an exquisite, multi-million-dollar Swiss Army knife. It is designed to survive contested airspace, gather sensor data, shoot down threats, and bring the pilot home. Because the pilot represents a priceless human asset and years of expensive training, the airframe must be built like a flying bank vault. Redundant flight controls, heavy life-support systems, ejection seats, armor plating, and complex thermal management systems add thousands of pounds of dead weight and billions of dollars in unit cost.
The entire pitch of the uncrewed Collaborative Combat Aircraft initiative is cheap mass. The promise is that we can build hundreds of these things for a fraction of the cost, swarm the enemy, and overwhelm their air defenses through sheer numbers.
Here is the dirty secret that nobody in uniform wants to say out loud: you cannot have it both ways.
If you pack an uncrewed airframe with the exquisite sensor suites, electronic warfare payloads, and high-G maneuvering capabilities required to match a crewed fighter, the price tag skyrockets. Suddenly, your expendable wingman costs thirty million dollars a copy. At that price point, nobody is going to hazard them in a high-attrition meat grinder against integrated air defense systems. You treat them with the exact same cautious, risk-averse operational mindset that you use for crewed jets. The supposed economic advantage evaporates the moment you demand fighter-level performance from a chassis that was supposed to be disposable.
Conversely, if you actually build them cheap—truly cheap, expendable attrition blocks—they cannot dogfight. They lack the kinematic performance, the power generation for heavy directed-energy or advanced radar jamming pods, and the structural integrity to sustain nine-G turns without shedding their wings.
The Air Force wants us to believe they have engineered a mechanical dogfighter. What they have actually done is institutionalized a semantic sleight of hand.
Why Autonomous Dogfighting Is A Dead End
Mention artificial intelligence in military circles, and someone will inevitably bring up DARPA’s AlphaDogfight trials, where an AI algorithm routinely pasted human F-16 pilots in simulated dogfights. It made for great headlines. It also proved precisely nothing about real-world aerial warfare.
Simulations are bounded playgrounds. The variables are clean, the physics engine is deterministic, and the enemy behaves according to programmed parameters. Real combat is messy, degraded, and fundamentally adversarial in ways a training loop can never fully capture.
Imagine a scenario where an uncrewed aircraft is operating in a high-threat corridor over the Western Pacific. GPS is jammed, satellite links are severed by kinetic anti-satellite weapons, and local communication nodes are drowning in a soup of high-power electronic noise. The aircraft is entirely on its own, relying on onboard neural networks to identify targets, assess rules of engagement, and execute lethal force.
Now introduce the edge cases that machine learning models hate most: adversarial data poisoning, unexpected atmospheric anomalies, or civilian commercial air traffic that looks superficially identical to a hostile asset. A human pilot brings millions of years of mammalian evolutionary heuristics to that cockpit. They can look at a canopy gleam, read subtle tactical postures, or make an intuitive moral judgment that
Why the Air Force Buying Uncrewed Fighters is a Massive Waste of Money
technology, news
The defense press is popping champagne over the latest delivery of uncrewed fighter aircraft to the United States Air Force. The lazy consensus is that robotic wingmen represent a clean, inevitable break from traditional aviation, shifting the cockpit paradigm into a safer, more scalable future.
It is a comforting narrative for taxpayers who want to believe technology can out-engineer attrition. It is also completely wrong.
I have spent decades watching defense acquisition programs hemorrhage billions into shiny objects while ignoring the structural physics of modern combat. Buying half-baked autonomous jets right now is not a stroke of genius. It is an expensive insurance policy against a threat model we barely understand, built on an industrial base that cannot support high-intensity loss rates.
The Economics of Attrition Are Broken
The fundamental pitch for uncrewed combat aircraft rests on a seductive lie: affordability through expendability. The pitch goes that if a drone costs a fraction of a crewed multirole platform, commanders can afford to lose them in contested airspace without triggering a congressional inquiry or a pilot shortage.
Look closer at the actual price tags.
Modern collaborative combat aircraft programs are creeping steadily upward in unit cost. Once you stuff these airframes with low-observable materials, encrypted datalinks, electronic warfare suites, and sensor fusion payloads, they stop being disposable kites. They become multi-million-dollar exquisite systems wrapped in a cheaper fuselage.
If a drone costs twenty million dollars a copy, it is no longer expendable. It is just a pilotless target.
Military planners love to talk about mass. Real mass requires a supply chain that can manufacture thousands of units a month under sustained missile attack. The current defense procurement apparatus builds hand-crafted artisanal fighters in small batches. Shifting the control mechanism from a helmet to a ground station does not magically fix a broken manufacturing ecosystem.
Latency is the Great Equalizer
Proponents talk about artificial intelligence pilots as if autonomy is a solved problem. It is not.
Flying a routine patrol in restricted airspace is easy. Executing high-off-boresight missile evasion while managing electromagnetic spectrum denial against an integrated air defense system requires intuitive split-second problem solving. Current machine learning models are brittle. They fail catastrophically when introduced to novel inputs or adversarial spoofing they did not encounter during simulation training.
Human pilots bring something irreplaceable to the cockpit: decentralized, embodied judgment.
When a datalink drops—and in a peer conflict, it will drop instantly due to saturation jamming—an uncrewed platform must fall back on pre-programmed behavioral rules or localized autonomy. If the adversary shifts tactics, your expensive robotic wingman turns into an expensive, confused paperweight flying at Mach 1.
We are trading human adaptability for algorithmic fragility under the false banner of progress.
The Training Pipe Dream
Another favorite talking point among defense analysts is the ease of scaling pilot replacements. You do not need twenty years of flight school to train someone to click waypoints on a tactical display, right?
Wrong.
Commanding a swarm of semi-autonomous assets requires a cognitive load that rivals traditional cockpit management. The operator sitting in a container thousands of miles away is still fighting the fog of war, only with a severely degraded field of view and sensory disconnect. Fatigue manifests differently when you are staring at a multi-monitor display drinking cold coffee, but it manifests all the same.
Furthermore, shifting the operator to the ground creates an entirely new vulnerability vector. A cyberattack on a stateside command node can ground an entire fleet halfway across the planet without a single shot being fired into airspace.
What Actually Matters
If the Air Force wanted to build a credible deterrent, it would stop chasing the ghost of the uncrewed dogfighter and focus on radical dispersion, automated logistics, and cheap, attritable mass that relies on simple physics rather than complex software stacks.
We do not need fewer pilots. We need fewer delusions about how wars are actually won.
Stop buying expensive compromises to satisfy congressional districts. Build things that break easily, cost peanuts, and overwhelm defenses through sheer volume. Until then, these new uncrewed jets are just expensive monuments to a military industrial complex that refuses to learn from history.