Karpathy: The LLM capability perception gap is widening—75% of people have barely touched one
karpathy · x · 2026-10-12
Karpathy re-upped his April tweet arguing the gap in shared understanding of LLM capability is widening into a sharp funnel:
- 6B people (75%) have barely come in contact with LLMs at all.
- 1-2B (20%) are casual free-tier users who see "derpy chatbots"—a better Google search or writing aid—and many reasonably find little use for them.
- Meanwhile this year's state-of-the-art agentic models (OpenAI Codex, Claude Code) are a different world from the old free models whose fumbles (voice mode mishaps, hallucination memes) still shape most people's views.
His core point: people's judgments are anchored to outdated, low-tier exposure, and the funnel is only getting steeper.
Related event: Karpathy: The LLM capability perception gap is widening into a sharp funnel(5 posts)→
More from AGI Musings
- Midjourney founder: human brain and frontier models cost about the same energy per thought — DavidSHolz · 2026-10-12
- AI safety researcher sees anti-scam tools winning long term, slow adopters at risk — QuintinPope5 · 2026-10-12
- 'I'm tired of the present': a struggler's case for fast AI takeoff — Realistic_Stomach848 · 2026-10-12
- We're in a pre-monopoly AI market: the case for separating model labs from agent apps — Abject-Tomorrow-652 · 2026-10-12
- Luddites Were the First to Attack AI: The Punch-Card Loom Was Computing's Ancestor — soumitrashukla9 · 2026-10-12
- Everyone Wants to Play the Meta Level: Why Self-Improvement Always Beats Doing the Thing — louisvarge · 2026-10-12