China plans to create at least 50 AI standards by 2026. This goal is part of a new draft policy from the Ministry of Industry and Information Technology (MIIT), as reported by the South China Morning Post. The standards will cover a wide range of areas including large language models (LLMs) and semiconductors.
This move is part of China’s effort to compete with the U.S. in AI development. Earlier this year, Alibaba’s chairman, Joe Tsai, mentioned that China is at least two years behind the U.S. leaders like OpenAI and Google in AI.
The proposed standards will include guidelines for LLMs, which are the basis for AI services like ChatGPT. They will also address safety, governance, industrial uses, software, computing systems, data centers, and semiconductor requirements and testing.
According to MIIT, these standards will apply to at least 1,000 Chinese tech companies. The draft policy also states that China aims to help create at least 20 international AI standards.
MIIT’s policy identifies 12 key AI technologies, such as LLMs, natural language processing, computer vision, and machine learning. It also outlines four layers of China’s AI industry: the foundation (including computing power, algorithms, and data), the framework, the model, and applications.
An expert mentioned that this draft policy is more market-friendly and innovation-oriented compared to usual regulations. This approach aims to support and grow China’s AI ecosystem, benefiting other industries too.
China’s tech giants, led by Huawei, are pushing forward in AI. Huawei’s second-generation AI chip, “Ascend 910B,” aims to compete with NVIDIA’s A100. Despite challenges in production, with a yield rate of about 20%, Huawei continues to advance.
In response to U.S. export bans, NVIDIA started selling its H20 AI chip tailored for the Chinese market earlier this year.
Source: trendforce