Andrej Karpathy Says only 5,000 People Have Seen What AI Can Really Do
Karpathy said about 5,000 people with internal access are seeing AI systems coordinate thousands of agents on software, cybersecurity and math work.
- On Sunday, former OpenAI researcher Andrej Karpathy wrote on X that the AI capability gap is "widening," describing the current state as a "sharp funnel" rather than two groups speaking past each other.
- According to Karpathy's napkin math, roughly 6 billion people have barely interacted with LLMs, while about 20 million professionals use frontier models for coding and math. Around 5,000 people maintain internal access to advanced systems.
- Advanced AI systems can now coordinate thousands of agents on massive software projects, but human review and comprehension are starting to fall behind. Karpathy warned the pace of advancement is so rapid professionals struggle to keep up.
- Box CEO Aaron Levie noted this gap illustrates how early we are in AI diffusion, as most public usage approximates next-generation search. He expects "a huge amount of change management" before industries can fully leverage autonomous agents.
- Companies still face challenges making these systems affordable, secure, and reliable for scale. Before automating document-heavy processes, businesses must connect data and address security concerns to ensure reliability.
12 Articles
12 Articles
AI researcher Andrei Karpathy, known as the creator of "vibecoding," points out that the gap is widening among people who use AI.
Andrej Karpathy, one of the founders of OpenAI, says there is a huge gap between users and developers, and very few people are using artificial intelligence for what it is actually intended for.
Anthropic's Andrej Karpathy says "the external world has seen the preview" of what AI can actually do
Andrej Karpathy, co-founder of OpenAI and now at Anthropic, sees a widening knowledge gap in AI. About 75 percent of the world's population has barely touched LLMs. Roughly 5,000 people, meanwhile, have access to internal frontier systems whose agent swarms do work that would take top experts years. The public, at best, sees a weird chatbot. The article Anthropic's Andrej Karpathy says "the external world has seen the preview" of what AI can act…
Coverage Details
Bias Distribution
- 67% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium
