It is either stored on the system and then processed hours or even days later when the machine returns from its mission. Or it is streamed back to an operations center in real time, terabyte by terabyte, which places a huge burden on military communications networks.
Either way, it is the job of humans, not machines, to comb through most of that data and find the relevant bits of information. In 2020, that is the full-time job of literally tens of thousands of members of the US military.
[…]
In reality, these supposedly “unmanned” systems require dozens of people to pilot each one remotely, steer its sensors, maintain it on the ground, and analyze the information that it collects, much of which is discarded because there are simply not enough people to process all of it.
Indeed, for years, the US military has supplied only a fraction of the drone missions that its commanders in combat have demanded. The problem has not been a lack of drones, but a lack of people.
[…]
In the absence of machines that can share information directly with other machines, this is how the United States connects its battle networks: a lot of people sitting in a lot of large rooms.
[…]
More often they use a computer-based instant messaging program called mIRC chat. I have watched individual servicemembers juggling a dozen separate chat windows, which can often involve taking information generated by one machine and manually transferring it to another machine. They call it “hand jamming” or “fat fingering.” It is slow and prone to human error.
A friend of mine who recently did targeting in the US military told me that the best way his unit could get on one page in identifying a target was with Google Maps. They had to gather up all of their different streams of information about the target from their assorted sensor platforms, come to a time-consuming decision on where the target actually was located, and literally drop a pin in Google Maps to direct their shooters where on earth to fire their weapons.