China sets out to ascertain a national standard for the Large Language Model (LLM). Plainly the federal government intends to self-develop LLM resembling ChatGPT and regulate artificial intelligence (AI) at the identical time.

The South China Morning Post (SCMP) announced on the World Artificial Intelligence Conference (WAIC) held in Shanghai by the China Institute of Electronic Standardization under the Ministry of Industry and Information Technology of China on the seventh that Huawei, Alibaba, Baidu, iFlytek, 360 to create national standards for LLM. , China Mobile and other firms have announced that they’ve formed a special task force.

That is the primary time China has launched a state-led LLM standardization task force. It might be read that the federal government will develop its own LLM that may compete with ChatGPT within the US.

Chinese firms are also speeding up the event of large-scale AI models.

Alibaba released ‘Tongyi Qianwen’, an LLM model much like ‘ChatGPT’ in April, and released ‘Tongyi Wanxiang’, a company image-generating AI, on the sixth. For those who input a prompt (command) in Chinese or English, it creates a sketch or 3D cartoon style image.

Huawei unveiled the three.0 model of the LLM ‘Pangu’, which was unveiled for the primary time in March. Pangu 3.0 features as much as 100 billion parameters, allowing users to coach AI models to suit their needs from different customers. Training through open data of a particular industry also can provide an AI model environment specific to the industry.

Baidu unveiled Ernie Bot in March, which, unlike ChatGPT, can perform mathematical calculations and generate images and videos with text prompts.

Meanwhile, the Web regulator China Cyberspace Administration (CAC) has yet to issue licenses for LLM products in China, despite tech giants resembling Baidu, Alibaba and Huawei launching ChatGPT-like services on pilots.

Li Yehong, CEO of Baidu, identified that while many players have entered China’s generative AI and LLM market, it is going to not be easy for them to acquire government approval, saying that there are ‘very high barriers to entry’.

China’s regulators announced draft regulations for services powered by LLMs resembling Erniebot, stating that the content they create have to be in step with the country’s core socialist values.

All LLMs and related products in China must undergo security testing and review by CAC before being released. “It’s the federal government’s goal to create national standards in order that AI will be trusted and controlled,” said CAC Director Zhuang Longwen.

Reporter Park Chan cpark@aitimes.com