Encouraged by the enthusiastic reception to its DeepSeek artificial intelligence platform in January, China's leaders are going all out to encourage AI companies to harness the power of this technology to compete with the United States and other countries in business and military spheres.
China considers AI an important tool to handle U.S. restrictions on Chinese business, particularly after DeepSeek shook up Wall Street, resulting in a loss of $589 billion for Nvidia stockholders in late January.
"The government in China works directly with the private sector and universities in the advancement and deployment of AI technology and are reducing their dependence on imports of high-technology products," said Lourdes Casanova, director of Cornell University's Emerging Markets Institute.
The past few weeks have seen China rolling out several new AI models, including Manus, which experts say can rival the latest model of ChatGPT. Industry experts were more than surprised to find that DeepSeek was equally efficient as ChatGPT, though it used older generation Nvidia chips. The U.S. has banned the supply of advanced chips.
"China and the U.S. have pulled way out front in the AI race. China used to be one to two years behind the U.S. Now, it is likely two to three months," Jeffrey Towson, owner of Beijing-based TechMoat Consulting, told VOA.
"Alibaba's Qwen is now a clear leader internationally in LLMs [large language models]. Chinese Kling AI and Minimax are arguably the global leaders in video generation," Towson said.
Government involvement
In 2017, China released an AI development program to make the country a world leader by 2030. The government's Next Generation Artificial Intelligence Development Plan said that AI would be adopted across different sectors and drive economic transformation.
"China has the most elaborate AI strategy compared to any other country," Rogier Creemers, assistant professor in Modern Chinese Studies at Leiden University in the Netherlands, told VOA.
China has established a National Computing Power Grid — somewhat like electricity grids — making it possible for Chinese AI companies to invest less in their own computing power. In the U.S., each company must fend for itself, Creemers said.
FILE - A Unitree H1 robot is displayed at a Unitree booth at the CES tech show Jan. 8, 2025, in Las Vegas.
Competition
ChatGPT's updated GPT4 large language model has gotten the attention of several top-ranking CEOs of Chinese tech companies. Baidu chief Robin Li recently said his firm was under "huge pressure and a sense of crisis" after seeing the updated ChatGPT. Baidu, which has launched Ernie Bot, said "the gap [between China] and leading international levels [in the field] has widened."
"AI plus robotics is likely where China will take a commanding lead over the U.S., just like in EVs,” Towson said. “Chinese companies like Unitree are already pulling ahead. Watch for China to surprise everyone in personalized robots, industrial robots and speciality robots,” he said.
Communist Party control
Chinese President Xi Jinping recently convened a meeting with heads of private companies, including tech firms, calling on them to "show your talent" in overcoming challenges such as an economic slowdown and U.S. restrictions on Chinese business.
"There are discussions that the growth of large language models — the technology behind chatbots like DeepSeek and ChatGPT — may be hindered by media censorship, because the models will have less diverse data to work with," said Creemers.
On the other hand, the government's control ensures industrial policy coordination, which is helpful in the growth of AI in China.
China is focusing more on specialized software for health and other industries, which can largely tolerate political censorship. Chinese AI models are improving diagnostic accuracy in diverse areas from detecting rib fractures to cancer.
SEE ALSO: Does AI detect breast cancer better than doctors can?US ban on advanced chips
"It will take some time, but it would not be a surprise if China is also soon capable of building advanced chips for AI," Cornell’s Casanova said.
Companies such as Huawei have shown that they can design and manufacture advanced chips successfully, thereby overcoming restrictions imposed by the U.S., she said.
Towson said China is 100% dedicated to building an independent semiconductor supply chain.
"It is advancing faster than anyone thought possible. But the frontier is always advancing, and it's unclear how this will play out over time," he said.
"But you can do a lot with software,” Creemers said. “China can work with more chips with less computing power or with fewer sophisticated chips."
The risk for China is not limited to chips, because the Trump administration could impose restrictions on the Chinese AI model. It could also react to China’s restriction on the use of ChatGPT, because it can violate its censorship rules.
SEE ALSO: China uses DeepSeek, other AI models, for surveillance and information attacks on USAI and the military
China's air force is using AI-powered biometric tests to screen potential pilots as part of a rigorous hiring process, according to state broadcaster CCTV.
"AI now plays a crucial role in interpreting candidates' biological signals, revealing underlying health risks that might not be immediately apparent to human evaluators," CCTV said. "This data-driven approach allows the air force to predict long-term risks, ultimately ensuring that only the most suitable candidates are chosen."
Chinese researchers have also revealed that the Chinese army has been using Meta's publicly available Llama model to develop an AI tool for potential military applications.