
The major language model of Deepseek-R1 can execute several tasks at a level comparable to the model created by Openai, the developer of ChatBot Chatgpt.Credit: Nicholas tocut via Getti/AFP
China’s technology startup DEEPSEEK has been swept the world of technology in the release of two large language models (LLMS), which is comparable to dominant tool performances developed by the US High Tech Jiants. It is built in just a small part of the power.
Scientists flock to DeepSeek: How to use the huge hit AI model
On January 20, Hangzhou, based in Hangzhou, is a partial part of Openai’s most advanced LLM, which is based in San Francisco, which can solve some scientific problems based on O1 Deepseek-R1, an open source “reasoning” model, has been released. California announced at the end of last year. And early this week, DeepSeek launched another model called JANUS-PRO-7B. This can generate images from Openai’s Dall-E 3 and stable diffusion prompts made by London’s stability AI.
If the performance of Deepseek-R1 surprises many people other than China, domestic researchers are expected to succeed in startups and become a global leader of artificial intelligence (AI). Say.
Many people, which are developing LLMS, and many people who have a doctorate in the field of science, technology, engineering, and mathematics, include Yunji Chen, AA AI, including AI Considering many people, it was inevitable that companies such as DeepSeek would appear in China. I am working on AI chips at the Computing Technology Research Institute in the Chinese Science Academy in Beijing. “Without DeepSeek, there will be other Chinese LLMs that can do something great.”
In fact, there is. On January 29, Tech Behemoth Alibaba has released the most advanced LLM QWEN2.5-Max so far. Last week, Moonshot AI and bytedance released the new inference model Kimi 1.5 and 1.5-PRO.
Government priority
In 2017, the Chinese government announced the intention of becoming an AI world leader by 2030. By 2025, IT was entrusted to the industry by completing the major AI break -through, “Technology and Applications Achieving the Leading Level of Technology.”
Developing a “AI talent” pipeline has become a priority. According to a report of the Security and Emerging Technology Center (CSET) at Washington DC’s Georger Town University, the Ministry of Education of China has approved a 440 university that provides AI specialized AI. According to Macropolo, Chicago, Illinois, China supplied almost half of the world’s leading AI researchers, and the United States accounted for only 18 %.
DeepSeek may probably benefit from investment in AI education and human resource development. This includes a large number of scholarships between academic and industry, research assistance, and partnerships, saying, Marina Zhang, a scientific policy researcher at Sydney University in Australia. China. For example, she adds thousands of initiatives, supported by states, such as the National Institute of Engineering, led by BAIDU in Beijing, have trained thousands of AI experts.
It is difficult to find accurate numbers about Deepseek’s labor, but the company founder Liang Wenfeng says that the company is looking for a graduate of a top -ranked university in China and a doctoral student student. Ta. Zhang says that some of the company’s leadership teams are under 35 and have witnessed China as a high -tech superpower. “They are deeply motivated by the willingness to independence in innovation.”
39 -year -old, Wenfen, graduated from the University of Zhijiang, a young entrepreneur and a major organization in Hangzhou. He jointly established hedge fund high flyers almost 10 years ago and established a deep seek in 2023.
Jacob Feldgoise, who is studying Chinese AI’s talents in CSET, states that national policy that promotes AI model development ecosystem will help companies such as DeepSeek in that it attracts both funds and talents. 。
However, despite the increase in AI courses at universities, Feldgoise has stated that it is not clear whether the number of students graduated from a dedicated AI degree and the skills that companies need are taught. Masu. In recent years, Chinese AI companies have complained that “graduates of these programs have not reached the quality they wanted,” and some companies have led some companies to partner with universities.