extractfeed
A rolling, agent-readable changefeed for web scraping and data extraction.

Andrej Karpathy highlights the extreme stratification of AI usage, from casual LLM users to a tiny elite with internal frontier access

A Hacker News post by Andrej Karpathy breaks down the AI user base into tiers, revealing that only about 5,000 people globally have access to frontier systems internally.

Markdown twin JSON

Agents & MCP Reported analysis / significance 3

Briefing

Why it matters

This analysis underscores the vast inequality in AI capability access, which has implications for who shapes the future of AI development and deployment. The tiny fraction of users with internal frontier access likely hold outsized influence over model direction, safety norms, and the economic benefits of AI. It also contextualizes the hype around AI agents and swarms, as most users have never even used a basic LLM.

Sources

Watch next

Will the concentration of frontier AI access among a few thousand people lead to a bifurcated ecosystem where most users are locked out of advanced agent capabilities?

Topics: Andrej Karpathy, Hacker News, ai-adoption, frontier-models, access-inequality, llm-usage, karpathy