The US Air Force has long been exploring the way artificial intelligence (AI) can assist in its operations — back in 2016 we saw combat AI beat some of the Air Force's top tactical experts, for example. Now, researchers are working on an autonomous drone designed to take down a piloted plane in air-to-air combat, with a showdown slated for July 2021.
As reported by Air Force Magazine, the Air Force Research Laboratory (AFRL) has been developing an autonomous fighter jet since 2018, with plans to use machine-learning technology in less advanced planes initially — such as the F-16 — before graduating to newer models, such as the F-22 or F-35.
Of course, details are a little hazy — it's a military project after all and therefore subject to hefty classification. In a video, however, head of the Pentagon's Joint Artificial Intelligence Center Lt. Gen. Jack Shanahan said the AFRL is "pushing the boundaries" of AI in a military application, adding that the team is quietly confident that the "machine [will] beat the human."
However, Shanahan also said that the Air Force would do well to heed the cautionary tales offered by other areas of autonomous innovation, noting that "There is no level four, full autonomous vehicle out on the roads today," and that as such, not all tech advances are out-and-out success stories.
Nonetheless, whether the project achieves its objectives or not, it joins a long list of other AI innovations being explored by the Air Force. One of the most notable, perhaps, is the Skyborg wingman drone project, which aims to create an AI-powered system of support for pilots in the sky. Elsewhere, the Air Force — and indeed US military overall — is already preparing to add machine learning to all areas of operations, from maintenance to combat strategy.