The Outlook for China’s AI Industry: Adoption and Applications

What are the key uses of AI in China, and are there any particularly innovative applications that have stood out to you recently?

Chinese AI diffusion has been going on for a long time, even prior to ChatGPT, which is how most people in the United States started to think about AI. As far as generative AI use cases, there are a ton. On the more frivolous end, you might see an AI avatar in Douyin or your favorite live stream e-commerce app selling nonstop for 24 hours. More serious use cases include using large language models and generative AI to enhance robotic performance in factories and supply chains. Additionally, AI adoption is much more prevalent in Chinese electric vehicles. Most electric vehicles in China have some level of an automated driving system ranging from automatic assistance to full self-driving. Robotaxis have been rolled out on a trial basis throughout important urban areas. Whether they’re innovative or not depends on your perspective, but at the end of the day, there is a much larger push for real-world use cases of AI and AI research as a part of the larger tech ecosystem in China.

Do applications of AI in China differ at all from applications of AI in the United States and other countries?

AI is a productivity enhancer. It can make things faster, easier, and more efficient, which are key characteristics of a general-purpose technology. Whether it is in China, the United States, Europe, or other parts of the world, AI applications are generally the same. For example, lawyers can become more productive by using the latest generative AI tool. The needs of a lawyer in the United States are not that different from the needs of a lawyer in China, even though the legal systems are different. In both countries, the application that has the most adoption is coding assistant tools. Code is being written more by AI, less by engineers. Every engineer or developer, whether they work in the United States or China or Brazil or France, is adopting these tools.

In what industries do we see the most AI applications?

The industries most prepared to adopt AI are industries with a lot of data reflecting the expected product. With coding, having a known “correct” answer and an abundance of coding data helps train AI tools to reach a software engineer’s level. The logic can expand to other use cases. For example, call centers are widely adopting AI because there are only so many ways you can respond to an angry customer, and recorded customer calls have been common for three or four decades. In China, manufacturing is another industry with a lot of AI applications. China has more factories than any other country around the world, with a lot of operational data collected for efficiency and quality. That data can be used to train whatever AI models are deployed inside factories. Overall, the prevalence of use cases is extremely dependent on data quantity and quality.

Are companies using AI mainly to increase efficiency, or are they incorporating AI for other novel and innovative uses?

From a company perspective, the focus for AI applications is unsurprisingly on enhancing profitability. There are two ways to increase profit: cutting costs or generating more revenue. AI is quite effective at cutting costs by increasing employee and system productivity. The more interesting angle is how companies are using AI to generate more revenue. For example, companies have used generative AI tools to produce a continuous AI avatar live streamer that is just as entertaining and can sell products just as effectively as a human. Right now, however, AI is primarily used for reducing costs and increasing efficiency. Soon, I expect there will be many more ways to use AI to generate new revenue.

How common is the use of AI for everyday purposes within China? Have there been efforts to bring AI outside urban tech hubs?

The average Chinese citizen’s attitude toward AI is much more optimistic than the average U.S. citizen’s. The Stanford Institute on Human Centered AI conducted a survey in 2022 that found that Chinese citizens are twice as optimistic as their American counterparts about the benefits of AI. The same survey was conducted in 2024 after the rise of ChatGPT and similar large language models. It found that optimism has increased a little in both countries, but Chinese citizens remain approximately twice as optimistic about the promise of AI as people in the United States. This difference in attitude means that Chinese people are more willing to try robotaxis, use self-driving systems, and interact with AI avatars and chatbots. AI is diffusing into daily life in China more quickly than in the United States. Of course, replacing search engines with AI is quite prevalent in both countries. But the higher optimism among ordinary Chinese citizens is noteworthy.

Why do you think people in China are more optimistic about AI?

China is an old civilization but a relatively new country, which is an interesting paradigm. It has experienced one of the most dramatic economic rises of recent human history, and a lot of that can only be done with a very open, and frankly risky, attitude toward adopting new technology. The past few generations of Chinese people have experienced more benefits than drawbacks from the adoption of new technology in areas ranging from high-speed rail to clean energy. This is not to say people do not think about the risks of technology. The Chinese government certainly has an opinion about what is good technology and what is bad technology, but as far as the everyday person’s concern, technology has been a net benefit, resulting in a more optimistic attitude toward whatever technology comes next, which just happens to be generative AI. Americans are more wary toward technology adoption, especially regarding data privacy. The same privacy concern exists in China, but the way ordinary people think about the trade-off is different.

Can you elaborate on how policies are affecting the development and adoption of AI applications?

When it comes to newer technology adoption, the Chinese government has typically had a more risk-seeking attitude in the sense that leaders are willing to let a certain technology develop in society first and regulate it later. This attitude is changing, however. China has introduced likely the most thorough and stringent registration system for AI. The model underneath every AI technology in China must be registered with the Chinese government before that service can be deployed. So now there is more of a “regulate first and deploy later” regime, which was not really the case in the 2010s during the heyday of China’s internet economy.

Generally, there is a willingness to let something happen in China when it comes to new technology, especially technologies the government finds attractive. This attitude boosts the country’s national productivity, hard power, and deep technological power. The deployment of robotaxis is an example of positive government intervention. The government designated experimental zones in major urban areas to allow robotaxis to improve before they were deployed onto the streets more widely. That kind of experimental zone does not exist on the same scale in the United States.

What opportunities and challenges do you foresee for the commercial application of AI in China?

The adoption curve in China has a similar set of challenges to other parts of the world. One is data, which is the fuel that goes into developing AI models. The model is just a mathematical abstraction of the data, and the collection of quality data could be very difficult for use cases like humanoid robotics. Because China is on the cutting edge of adopting and rolling out more humanoid robots, it will likely encounter that challenge before other countries.

China will also be the first country to have to deal with the social ramifications of labor disruption on a large scale. It will be interesting to see how the government responds to that challenge. Right now the government has a very go-go attitude toward deploying more robotics, more self-driving cars, and more automation to all walks of civilian life. At some point, there will be a large population of factory workers and drivers who are no longer needed. What will the Chinese government do with this population of people who need something to do? How will the rest of the world either learn or not learn from China? These are questions that we have not quite wrestled with in a very meaningful way, but will need to address sooner or later.

AI is a moving target. Whatever you hear today will change tomorrow—that’s what keeps things interesting. I hope more people will become optimistic about the opportunities of using AI for their own good without getting too discouraged by the challenges that are extensively reported in the news.

Kevin Xu is the founder and CIO of Interconnected Capital and author of the Interconnected newsletter.

This interview was conducted by Anna Xie while a Bridge to Asia Fellow with the Technology and Geoeconomic Affairs group at NBR.