Karpathy: a four-tier 'cognitive funnel' is splitting AI perception
Andrej Karpathy offered a "napkin math" estimate that roughly 75% of the world's population (about 6 billion people) has barely touched an LLM, while another 1–2 billion are occasional users of free ChatGPT-style products and therefore see AI as a "clumsy chatbot." He argues that some 20 million math and coding professionals are experiencing "agent vertigo" from paid frontier agents — especially OpenAI Codex and Claude Code — while at the very top only about 5,000 insiders have seen thousands of agents collaborate autonomously for weeks, hunt zero-days and push frontier mathematics. He attributes the split to three factors: the ambition and scale of your problems, the verifiable rewards available in code and math that suit reinforcement learning, and the access gap between free, paid and internal tiers. The population figures are his own estimates, not audited statistics.
It is a reminder that perceptions of AI capability depend heavily on a user's access tier, and that the public's 'AI looks dumb' experience and lab-internal experience can diverge sharply.