Which is one reason for the push to advance it without constraints.
I've been in to John Boyd's work lately. He was a big influencer in our militaries post-Vietnam maneuver theory of war. Derived from all his conceptual work on air combat.
The idea is to learn to read, react, decide, and execute at an operational tempo the enemy can't match. Like what we did in GW1. To get inside the enemy decision turn circle and st ay there so they are always stumbling back trying to get their footing. Once the starting gun fired we operated at a speed and tempo and flexibility that the enemy was powerless against. It was like they were frozen standing still compared to our 24\7 night\day agility. It was over before they even got a reply back from HQ on what their orders are.
AI would be the super weapon in that fight. It's ability to read, plan, research, simulate, decide, and execute would make any adversary who didn't have AI or who had weaker AI, appear to be as helpless Iraqis.
At some point we are going to have to simply turn control of weapons release over to them under certain circumstances. To utilize the true potential of AI it needs that freedom to act. If we hesitate and the enemy embraces that first, we will be helpless. It took a while for us to get comfortable with the idea of fly-by-wire. Fly-by-wire was really the embryo of AI. Putting a human in the loop will only bog down and prevent the technology from achieving it's full potential. Eventually someone is going to get over their squeamishness about that.
I know a couple of adversaries who would do that in a heartbeat if they thought it gave them the military edge.
