@krishnanrohit Yeah, I think it would be great to have an "office of net assessment for AI risk" that uses analogous methods to those for used by natsec community for assessing the risks from rogue/adversary states.
@BethMayBarnes I wonder if we could use the same capabilities view to think of other power-centers - like can Google or JP Morgan cause human permanent disempowerment? can North Korea? can the Pentagon?
Honestly not sure of the answer, and it would be a great thing for METR to publish.
