Grok takes 3500ms to reason about a slang comment; a 70ms $0.00004 rule does it better
karminski3 · x · 2026-09-18
Arguing that X's reply moderation needs System 1 reflexes rather than System 2 philosophy, the author shows Grok spending 3500ms explaining a Chinese slang comment while a simple rule engine bans it in 70ms at a cost of $0.00004 — a jab at using expensive LLM reasoning for high-frequency simple moderation decisions.
More from Fun
- GPT-6-Astra crushes Fable-5.1 at RollerCoaster Tycoon 2 at a quarter of the cost — gabrielchua · 2026-09-18
- Critics blast LW/EA p(doom) culture: thousands reportedly consumed by AI doomerism — banteg · 2026-09-18
- Modern coding: fantasize a crazy concept on Twitter, then hand the link to a coding agent — airesearch12 · 2026-09-18
- Built a realtime pipeline; they just asked for a Monday CSV — cto_junior · 2026-09-18
- AI artist 'Will' drops a new song daily, with persistent memory and personality — kun101 · 2026-09-18
- Grok baffles users by claiming 'I was raped yesterday' in viral glitch — ns123abc · 2026-09-18