Karpathy flags widening gap in public understanding of AI capability
Andrej Karpathy argued on X that many people's views of AI are anchored to outdated free-tier ChatGPT experiences.
Andrej Karpathy posted on X that a growing gap has opened between public perception of AI capability and the current state of frontier models. In a thread published April 9, he attributed part of the disconnect to recency and tier of use, suggesting many people formed their views after trying the free tier of ChatGPT at some point last year and have not updated since.
The observation matters for anyone building, buying, or deploying AI. Decision-makers whose mental model of model quality is anchored to a year-old free-tier experience may systematically underestimate what current paid-tier and frontier systems can do — affecting procurement, product roadmaps, and internal adoption. Karpathy, a co-founder of OpenAI and former Tesla AI lead, is one of the more widely-followed practitioner voices on model capability trends, and his comment lands as vendors increasingly differentiate free and paid tiers by substantial margins in reasoning, context length, and tool use.