Karpathy: Master the Model Before Building Agents
leslysandra · x · 2026-07-20
Andrej Karpathy points out that the biggest misconception in current AI development is forcing agents to work instead of thoroughly understanding the underlying model first. He emphasized that OpenAI made a similar mistake in 2016, costing them 5 years.
Core viewpoints:
- Fundamentals over appearance: Demos are easy to make, but turning them into products takes a decade, as proven by autonomous driving.
- Agents are not products: The underlying model is the true foundation. If you build a solid foundation, agentic capabilities will emerge naturally.
Related event: Karpathy: Master Base Models Before Pushing AI Agents(4 posts)→
More from AGI Musings
- FactoryAI’s Enoreyes says model distillation is basically unstoppable — LangChain · 2026-07-21
- Andrew Blumberg says formalization without interpretability is not science — AlexKontorovich · 2026-07-21
- Ken Ono says AI is forcing mathematicians to rethink how discovery works — soumitrashukla9 · 2026-07-21
- Open-source labs could distill a state-of-the-art model to 32GB or 80GB VRAM, the post argues — bookwormengr · 2026-07-21
- Two US companies are now using superintelligence to speed up the next generation of models — yacineMTB · 2026-07-21
- MIT Sloan says information, national security and finance are most exposed to AI — Exp_Mark · 2026-07-21