The Dangers of Artificial Intelligence

In August 2020, an artificial intelligence program defeated a human pilot in a series of 50 simulated dogfights. The AI was provided with information about its opponent's energy state that a true adversary would not have. If the skyborg decides it doesn't like an objective for a mission or something for some reason, or, you know, it goes off kilter,. It can then take out the human pilots that it's paired with. Could it do that? I mean, I don't know. There's real danger there. That's what freaks me out, maybe. Maybe I'm silly.