AI writing benchmark finds Claude Code replies use vague wording 60% of the time vs 4% for humans
IanArawjo · x · 2026-11-10
A benchmark study on AI 'slop' writing shows AI text routinely obscures main ideas and invents awkward phrasing, violating classic English writing principles far more often than human writing.
Key numbers from the accompanying chart:
- Vague wording: 60% for Claude Code replies and 25% for AI articles, vs just 4% for human articles and 7% for Stack Overflow answers
- Hidden characters and actions: 27% for Claude Code, 9% for AI articles, 1% for humans
- Empty statements and misplaced emphasis: 30% for Claude Code, 27% for AI articles, 4-6% for humans
Claude Code replies are the worst offenders overall. The thread walks through examples of the most common slop patterns.
More from Models
- Anthropic names its external alignment evaluators for Fable 5.2 — and one is a mystery — austinc3301 · 2026-11-10
- Polymarket gives OpenAI's next Astra model a 58% chance of shipping by Thursday — Polymarket · 2026-11-10
- Fable 5.2 is now live in Conductor — charlieholtz · 2026-11-10
- Dev builds an entire MMO world in one day with Fable 5.2, saying stylized art is the real test for new models — TAbrodi · 2026-11-10
- Fable 5.2's thinking trace shows 'answer thrashing' as it debates whether to dive into an art project — repligate · 2026-11-10
- Claude Fable 5.2 first tests: a messy playroom photo becomes a playable racing game on first try — every · 2026-11-10