How a U-2 Spy Plane Used AI and Made History
The U.S. Air Force’s legendary U-2 Dragon Lady spy plane made history as the first military plane to fly using an AI program to control key sensors and systems.
In a tweet, Will Roper, the U.S. Air Force’s Assistant Secretary of the Air Force for Acquisition, Technology and Logistics and so-called acquisition Tsar announced the mating of an AI pilot program with the U-2 Dragon Lady airframe, saying:
“NEW. For the first time, @usairforce put #AI safely in charge of a U.S. military system. Call sign “Artuμ,” we modified world-leading μZero gaming algorithms to operate the U-2’s radar. This first AI copilot even served as mission commander on its seminal training flight!”
The AI program, playfully called ARTUµ after the iconic droid helper R2-D2 from the Star Wars series, helped pilot a reconnaissance flight near Beale Air Force Base. During the flight, ARTUµ looked for ground-based missile launchers that could have posed a threat to the airframe, while the pilot kept an eye out for incoming enemy aircraft. Both the ARTUµ and the human pilot shared the U-2’s onboard radar, though ultimately ARTUµ decided to dedicate radar to missile detection.
Roper went on to explain what the melding of man and machine means for the future of flying, referencing sci-fi pop culture, saying, “Like any pilot, Artuμ (even the real R2-D2) has strengths and weaknesses. Understanding them to prep both humans and AI for a new era of algorithmic warfare is our next imperative step. We either become sci-fi or become history.”
An Important First
The flight marked the first publicly-known time that an artificially-intelligent program was involved with the flying of a military plane. Though not directly in control of the plane, the AI program controlled the airframe’s navigation as well as radar control sensors.
Prior to the flight, which took place in California, the program had successfully completed over one million training flights and is based on a gaming algorithm known as µZero, which has been previously used to best human opponents in popular and complex games like chess or Go, a strategy game popular in Asia.
In an interview, the pilot, identified only as callsign Vudu said that the program’s “role was very narrow … but, for the tasks the AI was presented with, it performed well,” though the human pilot remained “very much the pilot in command.”
Roper explained what the implications of this flight are for the future of AI flight and the United States military, saying that ARTUµ “was the mission commander, the final decision authority on the human-machine team. And given the high stakes of global AI, surpassing science fiction must become our military norm.” Stay tuned for more on ARTUµ, and for more on the future of artificially intelligent military flights.
Caleb Larson is a defense writer for the National Interest. He holds a Master of Public Policy and covers U.S. and Russian security, European defense issues, and German politics and culture.