@Yukon
May I point out that during the Heron systems AI in the test that you are (what I think you are) referring to was in conditions in which were unrealistic. Keep in mind that because in the test the only weapon allowed was the M61. Once the shot was fired with even a probability of being a successful shot, it was assumed to be successful. This leads to a very unrealistic scenario in which the AI was taking shots with much lower probability and this allowed for its quicker kill time. Banger was being more patient waiting for a higher probability shot. So I take the test with a grain of salt, the ai was facing very unrealistic probability with shot success. Additionally, the AI knew the data of bangers F-16 in the scenario, allowing for it to always know exactly what it was up against. In the real world such data is not what the AI would get.
The way I look at it is that in reality, banger may well have won the engagements. I would wait until a real world test. This test demonstrates that in a perfect environment with no adversity in knowledge about the opponent and a overblown shot probability, the AI is superior. But that is not the real world, and until I see the same results in the real world and dealing with probability and lack of knowledge of the opponent, I don’t see AI replacing pilots anytime soon. Loyal wingmen AI drones for strikes and Missile trucks, I can see them being implemented in the near future.
