As Models Master Structured Tasks, Creative Writing Keeps Getting Worse
teodorio · x · 2026-09-09
Responding to the puzzlement that 'AI can't write anything I want to read,' teodorio observes: the better models get at structured, finite tasks (math included), the worse they become at unstructured, open-ended ones like creative writing — and parts of code. Text specifically degrades as training keeps reinforcing previous quirks, explaining why capability gains and readability decline coexist.
More from Models
- DeepSeek reportedly running four V4 model versions, user says culling is coming to cut inference costs — teortaxesTex · 2026-09-09
- DeepSeek Accused of Silently Swapping Models Behind Its API, Sparking Debate — teortaxesTex · 2026-09-09
- Dev complains about over-blocking safety filters, asks Sam Altman for compute on Collatz — mertdumenci · 2026-09-09
- Dev claims turning off thinking mode makes AI models 5x faster, urges default non-thinking modes — Daniel_Farinax · 2026-09-09
- Yoav Goldberg: OpenAI's "cannot rule out" user data use is a lawyered admission — MannyKayy · 2026-09-09
- Heavy user on Astra: beats a junior hire on cost, but still fumbles simple tasks — RachelVT42 · 2026-09-09