but the argument in the OP is for replacing manned pilots with remote pilots, arguing that switch is ready now with existing tech. i stand by my assertion that hitting a stationary building on the ground is an entirely different thing than chasing, evading, or hitting another jet actively engaged in the same objective, and without this currently-non-existent (though possible) autonamous dogfight AI, the drones would lose.
I would ordinarily tend to agree with you but I rather suspect that AI software does already exist. And yes, we humans that game tend to be able to out think the silly AI in our FPS games so it would appear we will maintain an advantage. What tends to be forgotten is the fact that gaming developers tune the AI to be just challenging enough to keep us engaged without it being so overwhelming that we just give up and go play some other game. What if the developers of real-time combat AI do not have that constraint?