Earlier quoted context omitted.
As a thought experiment. How might an AI trying to follow the intent of Azimov's three laws of robotics help the humans? Initial conclusion: the community is broken and dangerous to humans. Evacuation of at risk life necessary. Distribute and re-integrate across functional settlements.
That conclusion might be adjusted a bit when the AI reads into our history, and observe that our track record for resettling people is not great.
Positive examples get much less coverage. Such as refugees that are spread out and integrated with host families and communities. Lives can be improved, and oversight and improvements might be had with the consultation or creation of experts in the subject.