Military procurement usually moves at a crawl. Bureaucracy swallows timelines. Budgets stall. Programs die quietly in committees. Yet when the U.S. Air Force Test Pilot School recently asked for three simple flight test concepts for its artificial intelligence research jet, the response caught everyone flat-footed. They expected a modest trickle of ideas. They got more than thirty.
The X-62 VISTA jet is turning heads for a reason. If you look past the standard military jargon, this modified F-16D is basically a flying laboratory where algorithms replace human muscle memory. When the 412th Test Wing hosted its first research symposium at Edwards Air Force Base in California, over eighty defense contractors, university researchers, and joint force representatives showed up. They didn't just attend. They flooded the command with proposals.
Why the sudden rush? Everyone wants a piece of aerial autonomy. The X-62, short for Variable In-flight Simulator Test Aircraft, proved its mettle back in 2023 during DARPA’s Air Combat Evolution program by pulling off combat-style dogfights against a crewed F-16. Since then, the aircraft has shifted from an experimental curiosity into the gold standard for testing machine learning in actual flight conditions.
The Air Force originally wanted just three concepts to fill out its fiscal 2027 and 2028 schedules. Instead, participants dropped over thirty detailed test pitches on the table. Col. Maryann Karlen, commandant of the Test Pilot School, admitted they were completely awed by the volume. The demand for an open testbed is massive, and traditional defense primes no longer hold a monopoly on good ideas.
Recent flight campaigns show just how fast things are moving. Take the HAVE HOLIDAYS series run by the test school at Edwards. These short, sharp missions were designed to see how quickly engineers could cram third-party software onto the jet and get it airborne safely.
During the HAVE BALLDROP test, the team integrated an autonomous agent from an outside vendor for the very first time, sourcing it through an Air Force Research Laboratory program called xFACTOR. Meanwhile, HAVE THANKS tested BirdBox, an open-architecture framework that helps onboard sensors recognize tactical targets at the edge.
Safety remains the ultimate bottleneck for military AI, which explains why HAVE DREIDEL is so important. That mission tested commercial runtime assurance software built by 3Laws Robotics. Think of it as an electronic leash. It acts as a hard set of guardrails designed to stop an autonomous agent from blowing past operational or safety limits established by human commanders.
Hardware is evolving just as fast. During the HAVE SANTA tests, engineers embedded IBM’s NorthPole chip straight into the VISTA's onboard computers. Modeled on biological brain structures, this low-power processor ran Lockheed Martin Skunk Works’ "Supermassive" simulation, crunching parallel flight trajectories mid-air without melting the avionics bay.
All of this happens while the X-62 undergoes a heavy Mission Systems Upgrade funded by the Test Resource Management Center. Lockheed Martin Skunk Works, Calspan Corporation, RTX, and the Air Force Research Laboratory are currently bolting on advanced radar, expanded computing power, and dedicated sensor bays. Once finished, the VISTA will tag-team with VENOM, a separate fleet of F-16s converted for autonomy testing down at Eglin Air Force Base in Florida.
Getting military pilots to trust machine learning isn't easy. Fighter pilots spend their entire careers mastering split-second instincts, so handing the stick over to a neural network goes against every survival instinct they possess. That is why briefings at the symposium included real talk from Air Combat Command fighter squadrons on how to build operator trust.
The defense tech landscape is shifting. Software startups and academic labs now have a literal runway to test algorithms that used to take a decade to clear safety boards. The sheer weight of those thirty-plus proposals proves the Pentagon doesn't have to beg for innovation anymore. The ideas are already waiting at the gate.
