But back to the article instead of your personal cheap shots. What I will take issue with is taking time to imagine some far off sci-fi world while the rest of us were in the real world practicing, fighting, studying and building TTPs for the foreseeable future so that we can integrate as a fighting asset in current/near-future contested environments (within limitations) or rescue package and as a reconnaissance/strike asset beyond the Army-S2-point-and-stare game. I've done actual contingency mission planning to bring unmanned to the fight in real time and for future OPLANs as well as plenty of lobbying with actual war planners, joint teams, and within AOC cells to leverage the very real skills that our unmanned assets possess. I’ve also cautioned when they're at their limits to steer the conversation back to getting unmanned assets into a viable role to achieve desired effects. And I think that's the key: useful thinking versus wishful thinking. What people ought to be reading are things like Squeeze Callahan's SAASS paper, because it can be taken seriously.
This article is fantastical and does little to prove that any of what he talks about can be done at the simplest level. For instance, the AOD/commander's intent isn't a 1's and 0's programming problem like he paints mission/targeting prioritization to be. We do it with preplanned ground targets and standoff weapons, but those typically don't move and the missile doesn't interpret commander's intent, the MPC and aircrew do.
And if a semi-autonomous system has to reach back to its operator at a critical juncture, it is vulnerable, whether from the increase in decision time thereby negating its computer advantage or from the transmission it just made to give itself away, negating its stealthiness.
Unmanned technology definitely has a future, but there are so many other problems to solve before frying his big fish.