Andrej Karpathy, a prominent artificial intelligence researcher who previously served as the head of AI at Tesla Inc. in between two stints at OpenAI, has announced plans to launch a new “AI-native” education startup.

The computer scientist and occasional educator revealed the launch of Eureka Labs LLC in a post on X today, describing his new venture as a kind of “AI-native school.” He explained that it will combine a traditional teacher with “AI symbiosis,” with human expert-created course materials scaled and guided via AI teaching assistants.

According to Karpathy, EurekaLabs represents the “culmination of my passion in both AI and education over ~2 decades… It’s still early days but I wanted to announce the company so that I can build publicly instead of keeping a secret that isn’t.”

⚡️ Excited to share that I am starting an AI+Education company called Eureka Labs.
The announcement:

—
We are Eureka Labs and we are building a new kind of school that is AI native.

How can we approach an ideal experience for learning something new? For example, in the case… pic.twitter.com/RHPkqdjB8R

— Andrej Karpathy (@karpathy) July 16, 2024

Karpathy shared his belief that for students, the ideal experience is to learn under the guidance of a subject matter expert that is “deeply passionate, great at teaching, infinitely patient and fluent in all of the world’s languages.”

But that ideal isn’t realistic for every student, because those kinds of experts are generally one of a kind. As such, there aren’t many of them and there definitely aren’t enough personally to tutor all 8 billion people in the world on demand, Karpathy said.

However, generative AI provides a potential opportunity to change that. With Eureka Labs, Karpathy envisages a future in which human teachers will still design the course materials, with AI teachers used to teach the actual courses. This will pave the way for an entire curriculum of courses to be made available on the same platform, Karpathy believes.

“If we are successful, it will be easy for anyone to learn anything, expanding education in both reach (a large number of people learning something) and extent (any one person learning a large amount of subjects, beyond what may be possible today unassisted),” he said.

To get the ball rolling, the company is planning to launch an AI course called LLM101n, which will teach students how to train their own AI model. Karpathy said the course materials will be made available online, but he also wants to run both digital and physical cohorts of students who can learn together.

In response to a question on X asking about his plans to monetize the company, Karpathy said he wants Eureka Labs to become a “self-sustaining business,” but at the same time, he doesn’t want to “gatekeep educational content” either. So that suggests some kind of business model where the basic course materials are available to everyone for free, with some kind of premium features or experiences.

Karpathy himself doesn’t appear to be quite sure yet, saying that Eureka Labs’ content will “most likely” be free and permissively licensed, and revenue derived from something else.

Rather than make money, it seems Karpathy’s ambition is to help everyone become more knowledgeable. “Eureka (from Ancient Greek εὕρηκα) is the awesome feeling of understanding something, of feeling it click,” he posted in another reply. “The goal here is to spark those moments in people’s minds.”

The reaction to Karpathy’s post was extremely positive, with dozens of replies praising his initiative. Eureka Lab’s X profile, which was only launched today, has already accumulated over 13,000 followers.

Industry analyst Holger Mueller of Constellation Research Inc. also reacted positively, noting that education is an extremely promising use case for AI because advances in the industry are desperately needed.

“The world is quickly transforming into a knowledge-based society, but the knowledge transmission process is little changed from the time when Aristotle gathered some kinds under an olive tree in Greece more than 2,000 years ago,” he pointed out.

Mueller said Karpathy makes a good case for AI in education, as the technology has already shown tons of promise in terms of creating text, images and video, which are all widely used for educational materials.

“Some of the more advanced multimodal capabilities of AI also lend themselves well to classroom environments,” Mueller continued. “For instance, an AI model can teach the same course to different types of learners and simultaneously ensure that those students are paying attention. So it’s fitting that Eureka Lab’s first educational course will be about AI itself.”

Karpathy said Eureka Labs represents the culmination of his passions for both AI and education. He is no stranger to teaching, having taught a computer vision course at Stanford University until 2015, when he departed to found OpenAI with Elon Musk, Sam Altman and others. He left the AI firm two years later to head up Tesla’s AI research team, helping to build the computer vision system that powers the Tesla Autopilot, an advanced driver assist system that relies on cameras to scan the car’s surroundings.

After leaving Tesla in 2022, Karpathy returned to OpenAI, where he helped develop and launch ChatGPT, which would go on to take the world by storm later that year, bringing global attention to the progress of the AI industry.

Karpathy left OpenAI for the second time earlier this year, and throughout his time there and at Tesla, he has remained an educator. For instance, he currently teaches an online course called Neural Networks: Zero to Hero, instructing beginners on how to build neural networks. In addition, he runs a YouTube channel where he occasionally posts lectures on AI-related topics.

Image: Andrej Karpathy