An AI-controlled F16 has performed its first ever dogfight with a human pilot, coming within 2,000 feet of each other at 1,200 miles per hour

pcgamer.com:

The rise of AI has many wide-ranging implications for the future of the human race, but it's perhaps the potential military uses that raise the biggest causes for concern. It seems we're now starting to see what an AI versus human battle might look like, at least in the skies above us, as the US military has tested an AI-controlled F16 in a dogfight with an actual test pilot—although the results as to who «won» remain unclear.

The catchily named X-62A Variable Stability In-flight Simulator Test Aircraft, or «VISTA» for short, is essentially a modified F16 fighter jet controlled by AI that has previously conducted multiple test flights to demonstrate the capabilities of its artificial pilot (via The Telegraph). It seems that testing has gone rather well, as last September the AI was let loose in the skies to test its dogfighting capabilities with real-life human opponents.

In a press release, the USAF Test Pilot School and DARPA revealed that they initially tested various defensive manoeuvres with the AI controlled jet to establish initial in-flight safety, before engaging in air-to-air simulated combat with another F16 in the skies above Edwards air force base in California last year. The AI-controlled aircraft and its traditionally piloted opponent came within 2,000 feet of each other in simulated combat, which sounds like a fair distance until you realise that they were travelling at 1,200 miles per hour. Eeep.

The AI algorithm analyses real-time data to make fast decisions in the air, allowing it to dogfight with real-life opponents and quickly respond to their reactions and manoeuvres in a way that mirrors the instincts of a trained fighter pilot. A series of safety protocols have been developed since initial testing to avoid in-air collisions and enable «human within-visual-range engagements», while human pilots remain on board with the ability to disengage the AI if it makes a critical error.

However, the Air Force says that its test pilots didn't have to use