Why China isn't leading the race for AI Supremacy

Kai-Fu Lee is a prominent figure in the field of artificial intelligence (AI). He has held senior positions at several major tech companies, including Google, Microsoft, and Apple. In addition to his work in the tech industry, Lee is an author and a leading AI researcher.

In 2018, in his book AI Superpowers, Kai-Fu Lee predicted that China would become the leader in Artificial Intelligence. He enumerated five reasons:

• Training AI models requires access to massive amounts of data, and because of its population, China generates a huge amount of data. Companies in China also have fewer restrictions on data collection than in some other countries, to the point that Kai-Fu Lee takes for granted that all this data is readily available.

• The Chinese government has made AI development a national priority, providing funding and support for research and development.

• According to Kai-Fu Lee, AI algorithms are pretty well known. You don't need discoverers as much as practitioners. China has a large and talented pool of engineers who can help create AI-powered startups and businesses.

• China has a thriving startup ecosystem. Companies are ruthlessly competitive, more than their US counterparts.

• Finally, Kai-Fu Lee notes that Chinese companies are often willing to take more risks than their US counterparts. This can give them an advantage in developing and refining AI applications, as they are more willing to experiment and try out new approaches.

Five years later, however, OpenAI and other US companies publicly lead the game. I think there are three reasons for this:

To train ChatGPT, OpenAI used text data from CommonCrawl, a large corpus of books, Wikipedia, news articles, and conversations (chat logs, social media posts, etc.), among other sources. It didn't need to access transaction data from people's shopping transactions and behavior—at least they haven't disclosed it. China's advantage in data collection may not be as significant as previously thought.

Although the algorithms are pretty well known, as Sam Altman explains in an interview with Lex Fridman which I've cited before, "the number of pieces that have to all come together, that we have to figure out—either new ideas, or just execute existing ideas really well—at every stage of this pipeline… There's quite a lot that goes into it." So yes, you need practitioners, but not any practitioners. It happens that OpenAI has managed to build a team of highly motivated, very talented people.

Finally, we cannot discard the influence of Sam Altman's leadership as CEO of OpenAI[^chinese-CEO]. I would bet that the same team, with different leadership, wouldn't have achieved the same extraordinary results. Companies succeed not only because of access to great resources, but thanks to exceptional leadership.

[^chinese-CEO]: If you want an example of the level of authonomy and decision-making the CEO of a Chinese company like TikTok has, check Show Zi Che's audience before the US congress and decide by yourself who is really running important decisions in TikTok.