AI takeoff arguments must count training and exploration compute separately
1a3orn and norvidstudies argue that takeoff analyses must separately account for the compute to train a model capable of algorithmic improvements and the compute for that model to actually explore and find them.
2026-11-22 ~ 2026-11-22 · 2 related posts
- Two compute costs that decide AI takeoff: training the improver and paying for its exploration — 1a3orn · 2026-11-22
- Takeoff math: separate training compute from the compute to actually explore — 1a3orn · 2026-11-22