aNdrey Kalpathy’s name is littered with the history of modern AI. In 2015, he was one of the founding members of OpenAI and worked as a research scientist. In 2017, he was personally hired by Elon Musk to lead Tesla’s computer vision efforts. He then returned to OpenAI in 2023 to work on improving GPT-4. But his greatest impact on the world may not be due to his research, but his role as one of the world’s leading educators on neural networks.
In 2015, Karpathy, along with his colleague Professor Fei-Fei Li, designed and served as lead instructor for Stanford University’s first dedicated deep learning course. Videos of his lectures have been viewed over 800,000 times. Since then, he has become one of the internet’s most beloved AI instructors. In recent years, Karpathy has explained complex topics to millions of YouTube viewers, such as how to build a smaller version of GPT from scratch.
“I’m a little obsessed with getting to the nitty-gritty of things,” says Karpathy. He attributes this to his physics education, which has made him adept at finding the simplest explanations for complex problems.
Karpathy draws inspiration from figures like Richard Feynman, known for his contributions to both research and public education. “People who are trying to learn a lot are clearly ahead of the money,” he says. “So I get paid when people appreciate me.”
His latest venture, Eureka Labs, was founded in July 2024 and aims to build “AI-native” schools. Its first product will be an undergraduate-level course on AI designed by humans and taught by AI teaching assistants. When he built his first AI course at Stanford nine years ago, he worked part-time and took months to complete it. “This course was a success,” he says. “But I think if we really focus on this, we can do much better.”
Purchase your copy of the TIME100 AI issue here
