So yes, the US Air Force really did put an AI in a modified F 16 and let it fly the jet over California with the humans onboard reduced to what is basically a very expensive witness statement. Frank Kendall went up in the X 62A VISTA at Edwards, in an aircraft that exists for one job: stress testing autonomy in a platform that normally relies on a flesh and bone pilot not making a single mistake while pulling G.
This was not a random PR lap either. It sits inside a longer run of work where the US has been pushing autonomy out of safe little sims and into real flight test points. DARPA has been blunt about what it achieved under Air Combat Evolution: AI algorithms flew an F 16 in air combat style scenarios against a human piloted F 16. The Air Force Research Lab has also publicly described the approach, including the reality that safety pilots are onboard with the ability to disengage if needed. That matters because the whole point is proving the machine can fly without the human having to catch it every five seconds. Here is the AFRL version, which reads like a calm press release while quietly describing a future where software updates become combat capability.
The jet is just the frame. The pilot was always the magic.
A lot of countries can buy shiny aircraft and park them on spotless aprons. The part they cannot mass produce is the person who can actually wring the full performance out of the machine at speed, under pressure, while being shot at, while a dozen systems are screaming for attention. Training produces a small number of truly sharp pilots, costs a fortune, takes years, and then you still have the problem that humans are fragile and capture is a national security event, not an HR issue.
An AI pilot does not black out. It does not get tired. It does not need oxygen. It does not panic, hesitate, or carry trauma around like a souvenir. It also does not care about dying, which is a polite way of saying it can be pushed into risk levels humans would never accept. If it gets taken out, you do not hold a funeral. You fix the model, load it again, and send another airframe. That changes the economics of air power in a way that is hard to overstate.
Now add the UK’s move on quantum navigation and the whole thing gets nastier.
Because the other weak point in modern aviation is reliance on satellite signals. GPS jamming and spoofing are not theory and they are not rare. It is common enough that civil aviation bodies have had to publish guidance and keep updating it as the interference spreads around conflict zones and beyond. EASA has been warning operators for years, and it keeps revising its material as the problem evolves. Here is an EASA update that spells out what jamming and spoofing do to aircraft systems in the real world. The FAA has also built an entire public resource hub on GNSS interference because this is now a persistent safety problem, not an edge case.
The UK is basically saying fine, we will build navigation that does not need the satellite handshake.
In May 2024, UKRI announced commercial flight trials of quantum based navigation systems that cannot be jammed or spoofed, done with Infleqtion and partners including BAE Systems and QinetiQ, tested out of Boscombe Down. The government announcement is even more direct, calling it un jammable quantum tech and putting a minister onboard for the final flight. That is not subtle. That is a signal. Here is the Boscombe Down detail.
This is also not a one day stunt. Imperial has already tested a prototype quantum sensor with the Royal Navy, and the Navy itself published the collaboration in plain terms: GPS free navigation that is less vulnerable to jamming and imitation. Here is the Royal Navy page on the navigation system.
Now connect the dots properly.
AI pilots remove the human bottleneck. Quantum navigation reduces the satellite bottleneck. You end up with aircraft that can fly harder than humans, operate inside messier electronic environments, and keep tempo even when the signal landscape is being deliberately poisoned. That does not just improve performance. It changes doctrine. It changes how attrition gets tolerated. It changes what a pilot shortage even means.
This is where the Terminator jokes show up, because they are a comfy way to avoid the uncomfortable sentence.
An AI controlled fighter with navigation that shrugs off jamming compresses decision loops. It pushes engagement tempo towards machine speed. It turns “human in the loop” into a policy aspiration that the battlefield does not care about. Meanwhile every defence ministry says the same thing every time a scary capability appears. If we do not build it, someone else will. Then they build it.
The skies are changing. The pilot is becoming the backup system. The map is learning to navigate without asking permission from a satellite.
If you want more of this, subscribe below. I will keep sending the sharp bits, and I promise to only use the word revolutionary when something actually starts replacing humans in real time.