At what multiple of Chinchilla-optimal compute do lab researchers anthropomorphize models?
eigenron · x · 2026-10-09
A tongue-in-cheek industry quip: the more interesting question is at what multiplier of Chinchilla-optimal training data a frontier lab researcher starts anthropomorphizing the models—a jab at researchers getting emotionally attached to models.
More from Fun
- Someone Used ChatGPT Instead of an X-Ray to Check a Hand Injury — amplifiedamp · 2026-10-09
- ESM Protein Universe: 7.7M protein families from 6.8B sequences turned into a 16-ft art wall — ylecun · 2026-10-09
- Elliot Glazer slams AGI-bet critics who refuse wagers and never admit being wrong — NathanpmYoung · 2026-10-09
- Hiring one AI to save time, ending up managing twelve AIs and a committee — max_paperclips · 2026-10-09
- GFL2 model release meme: Chinese modders are the best free advertising — Promptmethus · 2026-10-09
- Claude Opus 5.5 turns a Fallout 2 disc into a first-person walker demo — Promptmethus · 2026-10-09