ML researcher: LLM docs cram 3-4 idioms per sentence, ruining readability
ZeeshanZiaML · x · 2026-09-03
ML researcher Zeeshan Zia criticizes how LLM-generated text overuses idioms and metaphors—often 3-4 per sentence in the docs he reads. Meant to clarify, they instead confuse: he reports spending an hour or two on a few pages and retaining nothing, arguing this writing style seriously hurts reading efficiency.
More from Models
- A Gemini Flash model reportedly tops the DeepSWE coding leaderboard — sunjiao123sun_ · 2026-09-03
- Hypothesis: the better LLMs get at coding, the worse their writing gets — kwangmoo_yi · 2026-09-03
- PINNACLE: Claude Fable 5.1 halves agent failure rate but costs $2.46 per correct answer — ryanshrout · 2026-09-03
- 5 more days of free MiniMax M3, M2.7, Music 3.0 and Speech 2.8 on GMI Cloud, plus $500 hackathon — MiniMax_AI · 2026-09-03
- Reddit user finds popular Qwen 27B finetune underperforms stock after thinking time cut — Brief-Effect9065 · 2026-09-03
- Fable 5.1 benchmarked against 8 rivals across coding, 3D, SVG and data-viz — arena · 2026-09-03