A user highlights AI oversight proposals for lab inspections and compute monitoring
A post describes AI 2040's Plan A as monitoring computing resources, ensuring they can be destroyed in an emergency and making research transparent.
TLDR
A thread highlights “whole lab inspections”: a proposal to give auditors from the US and China enough access to be confident AI development is paused. It also describes AI 2040, which presents Plan A and Plan S. According to the thread, Plan A includes monitoring computing resources, ensuring they can be destroyed in an emergency, and research transparency intended to ensure those resources are used to make AI safer, not to race.
Combined views
660
1 Source, first seen ago