US military letting artificial intelligence ‘Hivemind’ fly fighter jet

The U.S. military wants to fly aircraft piloted by artificial intelligence. A few months ago, it succeeded when an AI piloted a U.S. jet fighter over American airspace. It’s yet another sign artificial intelligence and autonomous aircraft are advancing at breakneck speeds.

The Pentagon already said the aerial dogfights of tomorrow will no doubt feature combined squadrons of manned and unmanned aircraft. If the U.S. hopes to win that future fight and maintain air dominance, the military needs civilian contractors to help develop the technology. In the 21st century, that means partnering with software companies to develop artificial intelligence as well.

Tate Buffington is in business development at one of those companies, Shield AI.

“We’re primarily a software company,” Buffington said. “So, what we’re trying to do is climb the aviation food chain.”

Buffington said that means making the company’s AI agent adaptable. While at the Navy League’s Sea Air Space 2023 exposition, Shield AI demonstrated just how adaptable its “Hivemind” can be. The artificial intelligence became the first autonomous pilot ever deployed in 2018.

DefenseNews talks with CEO, Shield AI about MartinUAV Acquisition and Changing Contested Battlefield Dynamics with V-BAT + Hivemind https://t.co/jGMo9URdLC #UAS #RPA #AI #Autonomy pic.twitter.com/DSAspoqrLH

— USI (@USIinc) August 1, 2021

“We’re trying to get that right, the very first time,” Buffington said. “With that, it can go on multiple platforms. It can go on different UAVs, different large fighter jets, things of that nature. So as long as we get the autonomy right at the beginning, we can scale that up and down and be somewhat platform agnostic.”

The Hivemind first mastered flying smaller quadcopters. Then, the artificial intelligence agent started flying VBATs, a vertical takeoff and landing unmanned aerial system. The Marines and Navy like the VBATs because they don’t take up a lot of space, they can carry all sorts of different sensors, and can easily land on a moving platform like the deck of a ship.

In December, Shield AI’s Hivemind took another leap forward in autonomous flight. The AI agent piloted an F-16 Fighting Falcon for 82 minutes. The bulk of that time was spent flying out of visual range of the AI’s command and control unit. According to Craig “Bluto” Baker, a retired brigadier general and Air Force fighter pilot, it was a watershed moment.

Our Air Combat Evolution (ACE) program F-16 #AI algorithms graduated from computer simulation to piloting an actual F-16 in less than 3 years. AI agents recently controlled a specialized @TestUsaf F-16 in flight: https://t.co/kZbQo0gY1A pic.twitter.com/7VzwCuhSGA

— DARPA (@DARPA) February 13, 2023

“Just to see the aircraft and really hear from the pilot, who was actually in the aircraft, let go of the controls, push a button, and the agents took over. Just to hear the testimony and see the aircraft actually accomplishing the mission, flying the maneuvers based on the context of the mission and performing it almost flawlessly, was a great feat to see. Again, a technological revolution,” Bluto said.

The Pentagon already said it wants pilots in the future flying with autonomous wingmen. In some cases, a swarm of drones like VBATs could be sent into a fight while their human controllers are kept out of harm’s way.

“Not only will you see those DOD customers with one VBAT, you’ll see them with three and five and seven and 10,” Buffington said. “If you talk about different payloads, different sensors, you’re talking about a very heterogeneous team.”

Shield AI said its Hivemind can also help the military solve another major issue: the ever-expanding pilot shortage.

“If you look at conflicts that may be arising and the pilot shortages that are out there and the time required to train folks, that creates quite a bit of a gap,” Bluto said. “That gap can be filled with these unmanned aircraft that are piloted by this AI, by this autonomy.”