Josh Thorsteinson 🔸 comments on Why AI Safety Needs a Centralized Plan—And What It Might Look Like