The “Three Heroes of AI” are all from Guangdong, so why didn’t they develop in Guangdong?

Recently I saw a statement on the Internet: China’s “Three AI Heroes” are all from Guangdong, but none of them started their businesses or developed in Guangdong.‍‍‍‍

The " Three Heroes refer to Liang Wenfeng, the founder of Deepseek, ‍‍‍Yang Zhilin, the founder of kimi, and He Kaiming, an outstanding scientist in the field of artificial intelligence.

Public information shows that these three world-renowned AI experts are all very young. Two of them were born in the 1980s and one was born in the 1990s. One went to college and started a business in Hangzhou, one went to college and started a business in Beijing, and the other taught at MIT in the United States.

01

Liang Wenfeng: From Zhanjiang, Guangdong‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍

Single-handedly crashed the U.S. stock market‍‍‍

Deepseek is a domestically produced large model that became popular all over the world during the Spring Festival. It successfully reduced the training cost of large models by a cliff. It was called "Pinduoduo in the AI ​​world" and shocked the world. Some media even said that it "collapsed the U.S. stock market single-handedly."

Liang Wenfeng is the founder of Deepseek. He was born in 1985 in a remote rural area of ​​Wuchuan, Zhanjiang, Guangdong. At the age of 17, he was admitted to the Department of Information and Electronic Engineering of Zhejiang University with excellent grades and obtained a master's degree from Zhejiang University.

After graduation, Liang Wenfeng did not choose to work as a programmer in a large company like many of his classmates. Instead, he chose to start his own business and engage in fully automatic quantitative trading, which was not popular at the time. This is to use mathematical models and computer methods to guide stock trading.

In 2015, Huanfang Quantitative, founded by Liang Wenfeng , was officially established. In 2019, its fund management scale exceeded 10 billion yuan. In 2021, Huanfang's asset management scale exceeded 100 billion yuan, ranking among the "Four Heavenly Kings" in the field of domestic quantitative private equity.‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍

It was not until 2023 that Liang Wenfeng entered the field of AI ‍‍‍and founded DeepSeek. In less than two years, he achieved breakthrough achievements in the field of AI.

It can be seen that although Liang Wenfeng was born in Guangdong, his most critical university experience and career are closely related to Hangzhou.‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍

02‍

Yang Zhilin: From Shantou, Guangdong

Founder of the domestic large model Kimi

Yang Zhilin is the founder of another very popular domestic large model - Kimi. He was born in Shantou, Guangdong in 1993. He was admitted to the Department of Thermal Engineering of Tsinghua University in 2011, but changed his major to the Computer Department in his sophomore year. He graduated with a Ph.D. from the School of Computer Science at Carnegie Mellon University in the United States .

During high school, Yang Zhilin, who had no programming foundation, was selected into the Informatics Olympiad training class. He later won the first prize in the Guangdong division of the National Youth Informatics Olympiad League, and thus obtained the qualification of being a recommended student at Tsinghua University.

In 2011, Yang Zhilin was admitted to the Department of Thermal Engineering at Tsinghua University, but changed his major to the Department of Computer Science in his sophomore year. Later, he obtained a Ph.D. from the School of Computer Science at Carnegie Mellon University in the United States. He has worked at Facebook Artificial Intelligence Research Institute and Google Brain Research Institute.‍‍

In March 2023, Yang Zhilin founded Dark Side of the Moon Technology Co., Ltd. in Beijing. In October of the same year, he released the large model Kimi, which quickly attracted widespread attention. The current financing amount has exceeded US$1 billion and the valuation has exceeded US$3 billion, making it a super unicorn in the domestic large model industry.‍‍‍‍‍‍‍

03‍

He Kaiming : From Guangzhou, Guangdong

The success of ChatGPT is inseparable from his contribution‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍

He Kaiming is a top scientist in the field of artificial intelligence. He was born in 1984 in a wealthy family in Guangzhou. Both of his parents worked in management positions in enterprises. In 2003, He Kaiming graduated from Guangzhou Zhixin Middle School and was the top scorer in the Guangdong Province College Entrance Examination .

From 2003 to 2007, He Kaiming studied in the Basic Science Class of the Department of Physics at Tsinghua University . After graduating from undergraduate studies, he entered the Chinese University of Hong Kong for graduate studies, where he studied under Tang Xiaoou, a famous scientist in the field of artificial intelligence. He has worked at Microsoft Research Asia and Facebook, and currently works at the Massachusetts Institute of Technology (MIT) in the United States.

As a scientist, He Kaiming's contribution to artificial intelligence is not the founding of a star enterprise, but is more reflected at the academic level.‍

How much has He Kaiming contributed to artificial intelligence? Technology media said that he has removed a dark cloud over artificial intelligence, and even the birth of Chatgpt is inseparable from his academic contribution.

As we all know, the core of the artificial intelligence revolution sweeping the world today is artificial neural networks and deep learning. Scientists have realized that by increasing the levels of artificial neural networks, higher image recognition accuracy can be achieved.

But this was soon proved to be "wishful thinking", because after a certain turning point, the increase in the number of neural network layers would cause the accuracy of neural network recognition to drop rapidly. This problem was called "deep network degradation", and artificial neural networks also stopped at 19 layers.

At the time, this was called a "dark cloud" hanging over artificial intelligence . Without solving this problem, it would be difficult to make further progress in the field of deep learning.

In 2015, ResNet (Deep Residual Network) created by He Kaiming's team came out, which effectively solved the degradation problem of deep neural networks during training.

Before ResNet, the limit of what humans could achieve was a 19-layer neural network, and after that, it leapt to 152 layers. ResNet also became an important foundation for Google's star products such as AlphaGo Zero and ChatGPT.

Because of this, He Kaiming has become a very key scientist in the field of global artificial intelligence.

Open Google Scholar and search for He Kaiming, the number displayed is "520,000", which is the number of citations of He Kaiming's papers. In the entire artificial intelligence community, it ranks third, second only to the two veteran predecessors known as the "fathers of deep learning" - Joshua Bengio and Geoffrey Hinton.

He Kaiming, an outstanding artificial intelligence scientist

04‍

Why didn’t anyone start a business in Guangdong?

It is related to education and business culture‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍

The above is the basic information of the “Three Heroes of AI”. They were all born in Guangdong, one was born in the provincial capital Guangzhou, one was born in Shantou in eastern Guangdong, and one was born in Zhanjiang in western Guangdong.

Guangdong is China's most populous province with a strong talent pool. It is normal for it to export several outstanding talents. But if you must ask: Why don't they start their own businesses in Guangdong?

I think that in addition to their personal development, there are two environmental factors that cannot be ignored . One is that Guangdong lacks world-class scientific research institutes and cannot accommodate world-class scientific research talents . Admittedly, although Guangdong also has many good universities, there is still a gap compared with Tsinghua University, Zhejiang University, and MIT.

Guangdong reaped the first dividends of reform and opening up, and at that time had the first-mover advantage. People could make money by simply opening a factory or engaging in trade. But times have changed now. All walks of life are crowded and competitive, and it is a red ocean. If you want to make money, you have to use technology.

Have you noticed that the most popular entrepreneurial projects these days are no longer the ones that the general public can participate in, but are almost all played by highly educated people. The knowledge threshold and IQ threshold for entrepreneurship are getting higher and higher.

To advance in science and technology, one must have culture and world-class universities as a support, and Guangdong suffers from its disadvantages in universities.‍‍‍‍‍‍‍‍

It is worth mentioning that the situation in California, USA, was once somewhat similar to that in Guangdong. Both are late-developing regions, far away from the country’s traditional cultural and educational centers, lacking good universities. They were once called a "cultural desert." However, with the construction of first-class universities such as Stanford University and California Institute of Technology, California successfully attracted the top talents and gave birth to Silicon Valley, thus changing the temperament of the entire west coast of the United States.‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍

Compared with Beijing and the Yangtze River Delta, Guangdong’s biggest shortcoming is education . If it can build a world-class university similar to the level of Tsinghua University and Stanford University, I believe the industrial structure, talent pool and cultural temperament of the entire Greater Bay Area will be able to reach a higher level.‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍

Another reason, I think, is related to business culture . Note that it is business culture, not business environment.

As a bridgehead for reform and opening up, Guangdong's business environment is absolutely first-class in the country. It is one of the most active regions of China's private economy, with a huge number of private economic entities, accounting for 96.48% of the total number of business entities.‍‍‍‍‍‍‍‍‍‍‍‍‍‍

Guangdong's business culture is also very distinctive, with "daring to be the first, pragmatic and innovative, open and compatible" as its core. This culture makes Guangdong's companies pragmatic rather than idealistic, flexible and agile, and able to quickly adapt to market changes.

For example, the clothing manufacturing industry in Guangdong has a short cycle from design, production to market launch, and quick capital recovery. This business culture of speed and quick profit has enabled Guangdong's business to achieve remarkable achievements in the past few decades.

Canton Fair Venue

However, this business culture that focuses on short-term profits is somewhat powerless when it comes to high-tech industries, which, especially in the field of artificial intelligence, often require long-term investment and patience.

For example, Huawei has invested tens of billions or even hundreds of billions of dollars in making HiSilicon chips, Amazon and Alibaba in cloud computing, Nvidia in making GPUs, and OpenAi in making large models, and they have been able to withstand the pressure of not making a profit for more than ten years.‍‍‍‍‍‍‍‍‍‍‍‍

These projects not only require the employment of a large number of highly educated talents, but also require many years of investment regardless of cost. This kind of vision and determination to sit on the bench is something that many traditional companies do not have.

Don’t just look at the success of individual cases like Kimi and Deepseek. There are many projects in this field that have invested a lot of money but ended in failure.

High-tech is really expensive and has persuaded many bosses to quit. However, if a company only focuses on short-term interests and quick returns, it will never be able to come up with projects like ChatGPT and DeepSeek.

Therefore, Guangdong's economy is very strong, but it is not very outstanding in fields that require long-term investment, such as semiconductors, integrated circuits, artificial intelligence, and biotechnology.

Guangdong's science and technology industry is very strong, but it is strong in technology transformation and To C product level, but it is not very outstanding in basic scientific research, original innovation and other aspects.

I believe that if Guangdong wants to achieve high-quality development, it must first attach importance to education . On the one hand, education can cultivate and attract more talents and form a cluster effect; on the other hand, it can also enhance people's vision and pattern, avoid short-sightedness and quick success, and be able to engage in projects with longer cycles and do bigger and longer-term business.

Therefore, there is a reason why Guangdong has been frantically setting up universities in recent years. This is really forced by the overall environment and trends.