AI novels win Prix Goncourt while AI papers get mocked: a native-speaker blind spot and Opus 5.5 slop
EliasEskin · x · 2026-10-22
Responding to a confusing timeline where half of people mock AI-generated papers while AI novels win literary prizes, EliasEskin offers two explanations: a native-speaker advantage (he read the Prix Goncourt excerpts without flagging them as slop, mistaking AI style for florid French), and Opus 5.5's own issues — taking its outputs verbatim without editing likely yields unreadable papers, something he expects the next model cycle to partially fix.
Related event: ICLR submission flagged 63% AI-generated, sparking academic debate(4 posts)→
More from Models
- Users complain Anthropic models have 'suddenly got really stupid' — gaganghotra_ · 2026-10-22
- AXA, Vaisala and others deploy NVIDIA Earth-2 open weather models in production — NVIDIA Blog · 2026-10-22
- The '4' in Gemini 4 was a randomly chosen number, insider says — prajdabre · 2026-10-22
- DeepSeek Search debuts at $0.75/1k searches, cheapest paid Search API, ranks 17th of 27 — ArtificialAnlys · 2026-10-22
- Lafa scales flow language models to 8B, matching AR quality at 2.5x throughput — AdtRaghunathan · 2026-10-22
- Jev 1.14 lands on Databricks, targeting fast decisions and complex tasks — matei_zaharia · 2026-10-22