The defense establishment loves a comfortable middle ground. It keeps acquisition budgets funded, flag officers employed on defense contractor boards, and aviation journalists writing breathless copy about manned and unmanned aircraft flying wingtip-to-wingtip in perfect harmony. The current gospel states that the next generation of air combat will not feature fully autonomous kill-boxes, but rather collaborative combat aircraft acting as dutiful mechanical sidekicks to human-piloted quarterbacks.
It is a comforting narrative. It suggests that while silicon is getting smarter, the fragile ego of the fighter pilot remains the irreplaceable centerpiece of American air power. You might also find this similar story useful: The Economics of Synthetic Foresight Deconstructing Mantic and Predictive Intelligence.
It is also entirely wrong.
I have watched defense acquisition programs bleed billions of dollars into transitional compromises designed to soothe institutional anxiety rather than win high-end state-on-state conflict. We are spending engineering cycles and flight-test hours trying to shoehorn artificial intelligence into a human-centric command structure that cannot process data at the speed of modern electronic warfare. If your combat aircraft requires a human in a lead jet to look out of a canopy, interpret a fused sensor picture, and radio targeting instructions to a drone teammate, you have already lost the latency war. As extensively documented in latest reports by ZDNet, the effects are significant.
Let us dismantle the lazy consensus piece by piece.
The Latency Limit of Human Sight
The core premise of the collaborative fighter model is that a human pilot provides moral authority, tactical intuition, and creative problem-solving that an algorithm cannot replicate. This sounds poetic during a congressional budget hearing. It is a suicide pact in a contested electromagnetic spectrum.
Modern aerial combat against a peer adversary like China or Russia will not be fought within visual range or even within standard beyond-visual-range missile engagement zones where human intuition matters. It will be fought at the speed of light through directional data links, frequency-hopping radio waveforms, and automated counter-jamming protocols.
Consider the physics. A human pilot sitting in a cockpit experiences severe physiological G-force limits, cognitive fatigue, and narrow visual fields. Information processing speed is bottlenecked by biological wetware. When a radar warning receiver lights up because a surface-to-air missile battery has painted your formation with an active-array fire-control radar, the time window between life and catastrophic destruction is measured in milliseconds.
If that human leader has to process the threat, evaluate the positioning of four loyal wingman drones, assign priority targets, and transmit digital authorization, the aircraft has already been reduced to a titanium expanding cloud.
Algorithms do not get tired. They do not experience spatial disorientation. They do not hesitate because of collateral damage hesitation algorithms or second-guess an electronic warfare cue. By forcing an autonomous platform to tether its tactical decision-making cycle to a human biological clock, we are intentionally neutering the primary advantage of machine intelligence: instantaneous, distributed, multi-domain processing.
The Economics of Expendability
Proponents of the manned-unmanned teaming concept argue that loyal wingmen are designed to be attritable—cheap enough to lose in high-attrition environments. But look closely at the engineering specifications coming out of current Collaborative Combat Aircraft programs.
We are loading these supposedly cheap drones down with sophisticated sensor suites, stealth shaping, internal weapons bays, and complex maintenance-heavy composite structures. Once you demand that a drone keep pace with a fifth-generation manned fighter, match its supercruise kinematics, and carry advanced electronic attack payloads, the unit cost balloons from millions to tens of millions of dollars.
When a drone costs thirty or forty million dollars, it is no longer attritable. It is a disposable asset that defense contractors will fight tooth and nail to protect, recreating the exact same risk-averse tactical mindset that governs our current fleet.
True autonomy changes the economic equation entirely. If you separate the sensing and weapon-release platform from the human entirely, you can optimize designs for specific, brutal missions. You do not need internal weapons bays if your low-observable drone is operating in a tight swarm where raw mass and electronic saturation overwhelm enemy air defense networks. You do not need expensive life-support systems, ejection seats, and canopy actuators.
Every ounce dedicated to keeping a human pilot alive in the formation is an ounce stolen from fuel fraction, sensor aperture size, or ordinance load.
The Training Bottleneck
There is another dirty secret the defense press ignores: we are running out of pilots, and no amount of drone babysitting will fix the pipeline.
Training a modern tactical aviator takes millions of dollars and years of specialized instruction. The pilot retention crisis in Western air forces is structural. Pilots are leaving for commercial airlines to escape endless administrative burdens, terrible quality of life, and bureaucratic stagnation.
If our future tactical doctrine relies on every manned fighter leading a pack of four drones, you have just multiplied the cognitive and tactical load on an already depleted pilot corps. Instead of focusing on basic tactical execution, our squadron commanders will be flying airborne data-link managers, babysitting autonomous software glitches, and troubleshooting network dropouts while trying to evade incoming long-range surface-to-air missiles.
We are taking our most expensive, scarcest resource—the human tactical aviator—and turning them into aerial IT administrators. It is an operational dead end.
The Correct Path Forward
The future belongs to cheap, distributed, autonomous mass operating under decentralized command-and-control frameworks.
This does not mean humans are entirely absent from the kill chain. It means humans belong where they belong: safely behind the lines in hardened command centers, or operating high-altitude airborne battle management nodes like modified transport aircraft or specialized command platforms, orchestrating theater-wide strategies rather than steering individual wingtips.
Imagine a scenario where a squadron of low-cost, expendable autonomous vectors floods an integrated air defense system from multiple vectors simultaneously. They communicate via decentralized mesh networks, dynamically reallocating targets based on real-time attrition and sensor updates without requiring a single radio transmission from a human pilot fifty miles away.
That is not a sidekick model. That is a paradigm shift.
Stop pretending that tethering artificial intelligence to a human ego is an innovation. It is an insurance policy for an obsolescent industrial complex that is terrified of making the human fighter pilot obsolete.
The next war will not be won by the pilot who flies beside a drone. It will be won by the side that realizes the pilot was the bottleneck all along.
