Lobsters12 min readintermediate
There's no point at which turning your brain off will work
Summary
Dan Luu argues that turning off your brain and letting LLMs write code unchecked doesn’t work; while agents can help, they still need human oversight, especially on out‑of‑distribution problems, and cannot replace skilled engineers for most production work.
- Assuming LLM output works without verification leads to broken software; human QA remains essential.
- LLMs still struggle with out-of-distribution tasks (obscure languages, board games), requiring supervision.
- "Meat proxy" loops can reduce some effort but rarely replace skilled engineers for high‑value work.
- Over‑promising LLM productivity often stems from selective examples; real projects still need extensive iteration.
Engineers evaluating fully automated code generation need to understand LLM limitations to avoid costly failures.
6/10
