Yudkowsky: pretraining is down to 7% of compute and spiky capability predictions are back
repligate · x · 2026-10-08
Eliezer Yudkowsky revisits his old prediction that AI capability would first spike in verifiable domains like coding and AI research. He admits the recent human-data-driven era made capabilities look less spiky, leading many to declare his scenarios falsified—but argues that era was transient. With pretraining now down to a 7% share of compute, he says trends have reverted to first-order theoretical predictions, criticizing those who over-updated on the latest month's news.
More from AGI Musings
- We used to celebrate the technologist — now AI has us celebrating the model — martyjbeard · 2026-10-08
- OpenAI releases 372 math results in one day, including an AI proof of the Unique Games Conjecture — LukeW · 2026-10-08
- LM4Sci Workshop at COLM'26 to Gather LLM-for-Science Researchers Across Anthropic, Princeton, UW — ChengleiSi · 2026-10-08
- Researcher says agent token spend tops his rent, courts frontier labs for access — iScienceLuvr · 2026-10-08
- Stanford Professor: Stop Spinning Negative Results, Brutally Cull Ideas That Miss Milestones — anshulkundaje · 2026-10-08
- OSHBuilt's bet: MCP servers become the standard interface that lets AI agents source US manufacturing — hudzah · 2026-10-08