LLMs have read too many Medium articles: they always reach for the basic example APIs
_Stocko_ · x · 2026-09-21
The author quips that LLMs have read too many Medium articles — when writing code they always reach for the popular, basic example APIs first. A vivid observation on how training-data convergence pushes models toward clichéd solutions.
More from Fun
- Same-prompt test: Grok 4.7 takes 32 min at $8.14 while free SWE 2 finishes in 15 min — iamfakhrealam · 2026-09-22
- Steve Jobs' 1983 speech eerily predicted the iPhone, the Internet and AI — HeyAmit_ · 2026-09-22
- AI Twitter meme mashes Jevons paradox with LangChain's 'Noul' type — marktenenholtz · 2026-09-22
- LeCun adds 'machine intelligence' to the long list of failed AI renames — ylecun · 2026-09-22
- Researcher Susan Zhang blasts NVIDIA's output, blocks account in GitHub drama — suchenzang · 2026-09-22
- We built the most powerful NPCs ever, and now we're the ones running their quests — generativist · 2026-09-22