I have a bookmarks folder on X called "EA Cult" (for tracking effective altruism/x-risk conspiratorial beliefs), and boy, it's grown a lot over the last week.
I don't doubt the researchers' earnestness, but it would be helpful if p-doom claims were less vague about how AI (LLMs specifically) present realistic existential scenarios that don't involve human actors driving .
It can also be true that these earnest beliefs are convenient framing/evidence for the C-suite in DC.
Catastrophic risk and exestential urgency are great PR for corporations seeking regulatory capture.