China’s Vice-Minister for Industry and Information Technology, Xu Xiaolan, said in a keynote speech at the conference, which concludes this Saturday, that the country is moving to develop a complete AI value chain that covers “smart chips and algorithm frameworks to industry-specific large language models (LLMs)”.
She said the Chinese government will increase its support for the domestic AI industry, which is estimated to consist of more 4,300 companies at present.
Echoing that message, Shanghai’s Communist Party Secretary, Chen Jining, said in his presentation that the financial metropolis is already encouraging AI “algorithm innovation, chip development and language data set advances”. He added that Shanghai has been successful in attracting new AI businesses and talent.
In its Annual Report on the Development of New-Generation Artificial Intelligence (2022-2023), the China Economic Information Service – a professional organisation under the state-run Xinhua News Agency – said the country’s AI technology sector has already entered the global top tier, making up around 16 per cent of companies worldwide that are involved in the technology.
An op-ed piece published last month by the People’s Daily, the official newspaper of the Central Committee of China’s Communist Party, suggested that generative AI is on its way to becoming an essential part of the workplace and people’s everyday life, which would lead to profound changes in society, stimulate economic growth and drive a new industrial revolution.
Generative AI refers to the algorithms, such as those that power ChatGPT and similar services, that can be used to create new content, including audio, code, images, text, simulations and videos.
The People’s Daily piece also said Chinese institutions have so far launched at least 79 LLMs with more than 1 billion parameters, a measure of the size and complexity of a model.
LLMs are deep-learning AI algorithms that can recognise, summarise, translate, predict and generate content using very large data sets. These represent the technology used to train AI chatbots like OpenAI’s ChatGPT.
Still, the views expressed by the Ministry of Industry and Information Technology’s Xu and Shanghai Communist Party leader Chen did not elaborate on the difficulties being experienced by the domestic AI industry because of US sanctions.
Huawei Technologies deputy chairman Ken Hu Houkun, who also serves as a rotating chairman at the Shenzhen-based tech giant, said the company is trying to solve the current “computing power” bottleneck that is impeding AI development in China.
The US government last August imposed a ban on the export to China of certain products from chip suppliers Advanced Micro Devices and Nvidia Corp, which has a near monopoly on graphics processing units (GPUs) used to train AI systems. Those restrictions have been hindering the country’s progress in various cutting-edge applications, from LLM development to autonomous driving.
Still, Hu indicated that Huawei has helped power the development of about half of existing Chinese-developed LLMs through the firm’s Ascend data centre solution, cluster networking for large-scale AI model training.
SenseTime co-founder Sean Tang Xiao’ou, meanwhile, said in his presentation at the WAIC opening that Chinese scholars have contributed to the global development of machine-learning technology and large AI models.
China’s AI ambitions could face a fresh blow as the US government is said to be considering restricting Chinese companies’ access to American cloud computing services, which would stop Amazon Web Services and Microsoft Corp from using the power of advanced AI chips to benefit their mainland clients, according to a report by The Wall Street Journal on Tuesday.
The potential US ban on Chinese firms’ access to American cloud computing services could turn out to be the next big hindrance for the country’s LLM development efforts, according to an industry executive at WAIC who declined to be identified. He said training AI systems requires significant computing power.
China’s massive demand for advanced semiconductors to power new AI development projects has already created a fast-growing market for smuggled GPUs, such as the A100 and H100 devices from Nvidia.