Why AI progress accelerated: Claude 4.5 kicked off narrow RSI and open-weight catch-up

maksym_andr · x · 2026-09-23

Extending his narrow-RSI thesis, the author explains the 9-month acceleration in AI progress: when people realized over the Christmas break how good Claude 4.5 was, narrow RSI effectively began. The same logic explains why open-weight providers caught up so fast — access to a very capable proprietary model for inference optimization, RL environment generation, data curation, experiment code, and training runs speeds up frontier-chasing, with or without direct CoT distillation.

Original post →

More from AGI Musings

AGI Musings channel →