Late to this discussion after a brief hiatus from this platform, but one of the more seemingly concerning takeaways from the rogue AI postmortems is that is that the experts tasked to study these incidents say behavior of AI "swarms" are at risk of becoming indecipherable