According to state-run research institutes, Chinese companies created 79 large-language models (LLMs) to develop artificial intelligence (A.I.) algorithms over the previous three years.
According to a report by research institutes under the Ministry of Science and Technology, LLMs trained using deep learning on enormous text data entered an “accelerated” phase in 2020.
According to the Sunday report, Chinese organizations released 2 LLMs in 2020 and 11 in the U.S. in 2021; each country released 30 LLMs.
The research, co-authored by the Institute of Scientific and Technical Information of China, found that U.S. organizations released 37 LLMs the following year, compared to China’s 28.
China leads this year with 19 LLMs to the U.S. 18.
“Judging from the distribution of large-language models released around the world, China and the United States lead by a big margin, accounting for more than 80% of the global total,” the press announcement stated. “The U.S. has always led the world in large-language models.”
U.S.-led export bans prevent Chinese companies from accessing semiconductors necessary to train LLMs and other complex computing jobs, threatening the country’s A.I. industry.
The survey examined China’s 79 LLMs and found that while 14 provinces and regions had created such technology, academic-industry cooperative development projects were “insufficient.”
After OpenAI released ChatGPT, Chinese tech titans, including Alibaba (9988. H.K.), surveillance business SenseTime (0020. H.K.), and search engine Baidu (9888. H.K.) built their chatbots powered by generative A.I. and LLMs.