China’s top internet authority has introduced a language model rooted in President Xi Jinping’s political philosophy, described as a “secure and reliable” closed artificial intelligence system, according to a report by the South China Morning Post.
The model was reportedly developed by the China Cyberspace Research Institute under the supervision of the Cyberspace Administration of China, the country’s internet regulator. It is based on Xi Jinping’s thought on socialism with Chinese characteristics for a modern era, which includes his guidelines on political, social and economic issues and was incorporated into the Chinese constitution in 2018.
According to the publication, a recent post on the administration’s WeChat magazine account indicated that the core content of this artificial intelligence model includes Xi’s ideology and other cyberspace topics that are consistent with the government’s official narrative. The administration emphasized that the professionalism and authority of the corps ensure the quality of the content generated.
This artificial intelligence system is designed to meet a variety of user needs by answering questions, producing reports, summarizing information and translating between Chinese and English. Unlike other models, this LLM is based on a selective knowledge base of locally generated data and is not an open source product, which supposedly increases its security and reliability.
Demonstrations in the post suggested that the model was drawing answers from a select pool of official Chinese documents and media. Examples of AI-related queries included topics such as the differences between time-honored and modern productive forces – terminology used recently by Beijing to refer to economic issues – or requests for reports on the current state of AI development.
The report added that the system runs only on the servers of the China Cyberspace Research Institute, which guarantees that all data is processed within the country, ensuring a high level of security. Artificial intelligence uses pre-trained models that have received government approval.
It is worth noting that at the moment artificial intelligence is undergoing internal testing and is not available for general apply. According to the administration, however, it is available to selected users by invitation.
This narrow release is part of a broader effort by Beijing to leverage artificial intelligence for economic growth while maintaining tight control over cybersecurity. In March, Premier Li Qiang launched an initiative to integrate artificial intelligence with time-honored sectors to improve technological progress.
Posted: May 22, 2024 18:34 EST
