news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

Zhidongxi (public account:zhidxcom）
authorvanilla
editLi Shuiqing

Zhidongxi reported on July 17 that early this morning,Andrej Karpathy, co-founder of OpenAI and former Tesla AI directorBig news: He founded aAI+Education Company Eureka Labs。

▲Karpathy officially announced the establishment of a new company

Karpathy defines Eureka Labs as a new type of “AI Native School”, will tutor students through the symbiotic relationship of “teachers + AI”. He also spoiled the company’s first product -AI Course LLM101n, which will be an undergraduate course that guides students in training AI models.

Nine years ago, Karpathy, who was still a doctoral student at Stanford, joined OpenAI as a co-founder. He went to Tesla to study autonomous driving for five years, and returned to OpenAI last year to pursue AGI. In February this year, he once again announced his resignation and revealed that he planned to work on personal projects.

Now, this "personal project" has finally come to an end.

1. “Teachers+AISymbiosis, courses may be free and open

Karpathy sees this new company as the result of his passion for AI and education for more than 20 years. He said that his interest in education drove him from teaching people to play Rubik's Cube online, to designing and teaching Stanford's first deep learning course CS231n, to the recently launched AI from scratch course; and his interest in AI brought him from academic research to Tesla's product development, and then to OpenAI's AGI research.

So far, he has considered all of the above jobs as part-time jobs, side jobs to his "real job". This shows his passion for AI education.

▲Karpathy officially announced the establishment of a new company

In the official announcement, Karpathy talked about the pain points of learning new knowledge: subject matter experts who are excellent teachers, patient, and proficient in all the world's languages ​​are very scarce and cannot personally tutor 8 billion people around the world.

The emergence of generative AI has brought new possibilities to this learning experience. With the help of AI teaching assistants, teachers can design course materials more easily, and AI can also assist students in completing courses. This symbiotic relationship of "teacher + AI" can effectively expand the scope of education.

Karpathy revealed that Eureka Labs' first product, LLM101n, will be an undergraduate course that guides students to train their own AI, similar to a scaled-down AI Agent. The course materials will be provided online, and there are plans to organize online and offline teams to study together.

Some netizens asked whether Eureka Labs' products would be provided for free or as a subscription service. Karpathy said that he hoped Eureka Labs could become a formal, self-financing enterprise. In his opinion, educational content itself should befreeYes, revenue will come from other aspects such as operating online or physical courses.

▲Netizens asked about Eureka Labs' business model

Industry leaders sent congratulatory messages. Jeff Dean, chief scientist at Google DeepMind, said: "Congratulations! AI will have a huge impact on education, and I know it's something you're passionate about (I saw how you organized CS231n when I gave a guest lecture a few years ago, and it's clear that you've put a lot of thought into studying the best ways to teach different types of students)."

▲Comments from Google DeepMind Chief Scientist

Emad Mostaque, former CEO of Stability AI, said: “Congratulations! Education is at the heart of everything and I can’t wait to see it go to the next level.”

▲Comments from former Stability AI CEO

2. He studied under Fei-Fei Li and had “double entry and double exit”OpenAI

Karpathy has an impressive academic and professional resume.

He studied for his undergraduate and master's degrees at the University of Toronto and the University of British Columbia, respectively. From 2011 to 2015, Karpathy studied for his doctorate at Stanford University under the tutelage of the famous AI scholar Fei-Fei Li, focusing on the study of convolutional/recurrent neural networks and their applications in computer vision, natural language processing and cross-disciplinary fields.

▲Karpathy’s academic experience

While studying, he had three internship experiences at Google Brain, Google Research, and DeepMind, participating in work related to video and deep reinforcement learning.

In 2015, he joined OpenAI as a research scientist as one of the founding members. In 2017, he was recruited by Musk to Tesla to lead the AI ​​and autonomous driving vision team.

During his five years at Tesla, Karpathy facilitated the development of Autopilot, an automatic assisted driving technology, which was crucial to Tesla's subsequent fully autonomous driving system FSD.

In February 2023, Karpathy returned to OpenAI in a high-profile manner, saying that he was "inspired by the impact the company had."

▲Karpathy’s work experience

A year later, he announced his resignation again, and said that his immediate plan was to "work on personal projects." Now, this "personal project" has finally come to an end.

It is worth mentioning that during the gap period after his two resignations, Karpathy has not been idle. He has uploaded 10 AI course videos on YouTube, and the cumulative number of views has reachedMore than 100 million times。

Conclusion: A powerful player has emerged in the AI+education track

AI has been implemented in thousands of industries, and education has always been a hot track. Huang Tiejun, chairman of the Beijing Zhiyuan Artificial Intelligence Research Institute, said at WAIC 2024 that he predicts that in the next few years, the cognitive and professional level of AI in the field of education will be comparable to that of university professors, doctors, and graduate students.

In China, the export of AI education products has become a major trend. Represented by Question.AI of Zuoyebang and Gauth of ByteDance, domestic applications have emerged as a new force in the overseas AI education application list. In terms of policy, the Ministry of Education issued four actions in March to promote the integration of teaching and learning with AI.

Today, Karpathy's joining has injected new vitality into this track. Although AI still has a lot of "illusion" problems, and there is a gap between its logical reasoning and deep understanding capabilities and those of humans, as an auxiliary tool, AI is showing its value in ensuring the accuracy and verifiability of proofs.