The development of LLM models, which are trained using AI deep learning techniques to mine huge amounts of data and text, has entered a "speed-up" phase since 2020, according to a report by research institutes under China's Ministry of Science and Technology.

The report released on Sunday said that in 2020, Chinese institutions issued just two LLMs compared to 11 in the US, but in 2021 there were a total of 30 LLMs issued in each country.

U.S. institutions issued a total of 37 LLMs in 2022, compared with 28 in China, according to the report. China has issued 19 LLMs so far this year, while the U.S. has added just 18 of these smart contracts.

“China and the United States lead the distribution of large language models worldwide, with more than 80% of the models being published by the two countries,” the report concluded. “The United States consistently leads the world in the number of large language models.”

The report comes as China’s AI industry faces significant challenges, with US-imposed export controls making it difficult for Chinese organizations to access semiconductors used to train LLMs, as well as other advanced computing tasks.

The report analyzed 79 LLMs developed in China, noting that although 14 provinces and regions have developed the technology, joint development projects have yet to “meet the demand.”

After OpenAI released ChatGPT, Chinese tech giants ranging from Alibaba to surveillance company Sensetime and search engine giant Baidu rolled out their own versions of chatbots powered by general AI technologies.

Hoang Ton (according to Reuters)