The United States Air Force flew a U-2 spy plane with an AI system working as a co-pilot with a human crew member. (Image Credit: Airman 1st Class Luis A. Ruiz-Vazquez/US Air Force)
For the first time, artificial intelligence controlled a U.S. Air Force military system. On December 15th, the United States Air Force flew a U-2 spy plane integrated with an AI system working alongside a human pilot at Bale Air Force Base in California.
The AI system, called ARTUµ ARTUµ, was responsible for managing sensor employment and tactical navigation while the pilot flew the plane and coordinated with AI on sensor operation. The pilot, U.S. Air Force Major Vudu, and ARTUµ participated in a reconnaissance mission during a simulated missile strike. ARTUµ was tasked with detecting enemy launchers while Vudu watched for threatening aircraft, both using the U-2’s radar. With no pilot override, ARTUµ determined when the radar should focus on missile hunting and self-protection. As a result, ARTUµ was the mission leader, but it only concluded where the pilot should fly.
ARTUµ took over sensor control once the U-2 spy plane took off. From there, the AI system manipulated the sensor based on the insight it learned from over 500,000 simulated training iterations. The pilot and AI system successfully partnered up to share the sensor and complete the mission.
The AI system, called ARTUµ, managed sensor employment and tactical navigation while the pilot flew the U-2 spy plane and coordinated with AI on sensor operations. (Image Credit: Airman 1st Class Luis A. Ruiz-Vazquez/US Air Force)
“ARTUµ’s groundbreaking flight culminates our three-year journey to becoming a digital force,” said Dr. William Roper, assistant secretary of the Air Force for acquisition, technology, and logistics. “Putting AI safely in command of a U.S. military system for the first time ushers in a new age of human-machine teaming and algorithmic competition. Failing to realize AI’s full potential will mean ceding decision advantage to our adversaries.”
Will Roper says that the team trained µZero gaming algorithms to operate the plane. The algorithm dominates games like chess, Go, and video games with no knowledge of their rules.
A million training simulations performed in over a month by the Federal Laboratory team went into μZero’s algorithm to operate the radar. It was taught how to detect enemies and search for danger while interacting with a pilot. The AI’s access to other subsystems was cut off to minimize the risk involved with its decisions. If it was navigating in enemy territory, the on-off switches could be adjusted.
The AI technology was designed so that it could be transferred to other systems. “Blending expertise of a pilot with capabilities of machine learning, this historic flight directly answers the National Defense Strategy’s call to invest in autonomous systems,” said Secretary of the Air Force Barbara Barrett. “Innovations in artificial intelligence will transform both the air and space domains.”
Have a story tip? Message me at: http://twitter.com/Cabe_Atwell