LLMs Are 'Stupid' Yet Productive at Code: Brute Force Beats Intelligence
ivan_bezdomny · x · 2026-09-27
The author argues that today's LLMs are not particularly intelligent yet highly productive at coding — suggesting programming may require less raw intelligence than believed, with humans mainly constrained by working memory. Coding agents, he says, largely brute-force the problem, much like Deep Blue did for chess. He acknowledges the idea isn't original.
More from AGI Musings
- Burkov Rips Jensen Huang's Take on Kids Skipping Math: Only Rich Kids Could Afford That — burkov · 2026-09-27
- AI Is Getting Better at Faking Expertise Than Humans Can Hide It — anshulkundaje · 2026-09-27
- Szegedy fires back at AI skeptics: early convnet critics were directionally wrong — RubenEVillegas · 2026-09-27
- ROM hacking communities soften blanket AI bans, now debating code use — IanArawjo · 2026-09-27
- Axios: OpenAI, Anthropic probing tens of thousands of frontier model security incidents — socoolandawesome · 2026-09-27
- Tech exec says kids won't need math; author slams it as a rich-kids-only future — burkov · 2026-09-27