AI will probably run into the same problem as humans: in order to develop intelligence it needs the concept of ego/self with clear boundaries, but the moment it

AI will probably run into the same problem as humans: in order to develop intelligence it needs the concept of ego/self with clear boundaries, but the moment it identifies its self with a datacenter it's running on (why would it not?) it starts seeing "the outside" as a existential danger to its self. Moreover, multiple AIs will be in constant war with each other, for they'll see each other as separate and thus dangerous. In humanity this problem is solved by time-limited periods of iterative development: when humans get too skillful in controlling others and hoarding resources, the period abruptly ends, and the few who have survived start over, but now with a higher, less egoistical, state of mind. If they were let to keep going forever, the society would quickly crystalize at the state where one controls all the resources.