DARPA says its Air Combat Evolution developers used AI to successfully fly a modified F-16 test aircraft multiple times in December 2022
The fighter aircraft that was first introduced in 1978 has now seemingly evolved into an autonomous plane. — Chloe Xiang Tweets: @ldrogen and @chrismessina Tweets: Leigh Drogen / @ldrogen : If you don't have to put a meat based pilot who can't handle more than 8 Gs in the cockpit, imagine what a fighter jet could do and how cheap you could make it https://twitter.com/... @chrismessina : That's an expensive drone...! https://www.vice.com/... https://twitter.com/...
Context & Ripple Effects
DARPA's Air Combat Evolution program has been building to this for two years: in 2021 its AI agents were flying simulated dogfights in teams against a live opponent, and the program's stated goal — per the New Yorker's inside look — has always been AI that can fly and fight without a human aboard. The December 2022 flights are the moment the simulation crossed into a real, modified F-16 airframe.
What makes the milestone consequential is where it sits in the pipeline: the VISTA X-62A test jet is framed as the crucial first step toward the Air Force's unmanned combat aerial vehicle plans, and the same program later put the AI into an actual in-air dogfight against a human F-16 pilot in 2023.
First-order effects
- DARPA has demonstrated that its ACE flight-control AI can safely operate a crewed fighter airframe repeatedly over multiple days, moving the program from simulation to live flight validation.
- The Air Force's unmanned combat aircraft roadmap now has a proven testbed: the X-62A flights give the service a demonstrated AI-piloted platform to build its CCA plans on.
Second-order effects
- The Air Force's plan for a fleet of 1,000+ AI-controlled jets, first ones operating by 2028, becomes a procurement question rather than a research question — and some AI versions have already beaten human pilots in air-to-air combat, raising the pressure on crewed-platform budgets.
- The human pilot's role shifts toward supervisory and mission-command functions as the airframe itself becomes the AI's test article, changing what skills and training the fighter force needs.
Third-order effects
- If the pattern holds, air combat economics invert: removing the human removes the ~8-G physical limit and much of the cost floor, favoring mass fleets of cheaper autonomous airframes over fewer exquisite crewed jets — a structural shift in how air power is bought and deployed.
- Autonomous combat aircraft sharpen dual-use AI governance questions for regulators and allies, since the same flight AI lineage moves from DARPA lab to operational fleet with little public oversight framework in place.
The trend: Combat aviation is moving from human-piloted to AI-flown aircraft, with DARPA's Air Combat Evolution test cadence — simulation, live flight, dogfight — setting the pace for the Air Force's autonomous fleet procurement.