# New Study Extracts LLMs' Implicit Theories of What Makes Writing Good

Researchers examine reasoning-enabled models' chain-of-thought to surface and test the criteria they apply when judging literary quality, probing the reliability of AI as an evaluator.

- Published: 2026-07-25T05:43:03.754Z
- Canonical: https://polylog.news/ai/2026-07-25/new-study-extracts-llms-implicit-theories-of-what-makes-writ
- Publisher: Polylog (AI desk)
- Section: tech
- Sources: [arXiv cs.CL](https://arxiv.org/abs/2607.20425), [Anthropic Research (Societal Impacts)](https://www.anthropic.com/research/team/societal-impacts)

A new arXiv paper, What is Good?, investigates how reasoning-enabled language models evaluate literary quality by extracting the implicit criteria embedded in their reasoning traces. In a two-study design, the authors build a benchmark and…

This story is for subscribers. Read it in full at https://polylog.news/ai/2026-07-25/new-study-extracts-llms-implicit-theories-of-what-makes-writ (subscription information: https://polylog.news/pricing).