China's Cyberspace Research Institute has introduced XiBot, a large language model and conversational AI rooted in the ideas of President Xi Jinping. The AI model is currently undergoing internal testing and is not yet accessible to the public.
According to a post published on Monday on the WeChat account of the administration's magazine, Xi Jinping's philosophy, along with other selected cyberspace themes aligned with the official government narrative, make up the core content of the LLM.
The creation of Chat Xi-PT, as the AI is informally known, aligns with Beijing's ongoing efforts to propagate Xi's ideologies. When Xi delivers significant speeches, his words are promptly reproduced online, making the transition to a chatbot format a logical progression. The AI model is designed to provide accurate information based on Xi's documented thoughts, potentially reducing the risk of AI hallucinations—a common issue with generative AI models—since in China, Xi's words are considered authoritative and incontrovertible.
AI Built on Government Documents
The system was “deployed exclusively on the servers of the China Cyberspace Research Institute, where all data is processed locally, ensuring a high level of security” and that is exclusively based on pre-trained, government-approved language models.
XiBot is constructed using a comprehensive amount of government documents, enabling users to query these documents through a generative AI interface. The ideological corpus of Xi Jingpin´s thought, officially known as “Xi Jinping Thought on Socialism with Chinese Characteristics for a New Era,” includes his instructions on all aspects of political, social, and economic life. It was enshrined in China's constitution in 2018.
This book is widely available in China and is often seen in airports frequented by Chinese tourists or members of the Chinese diaspora, translated into various local languages. Xi Jinping's thoughts are also extensively featured on Chinese government websites and news outlets, ensuring his philosophies are well-disseminated both domestically and internationally.
Potential Implications and Comparisons
Creating AI models based on prominent figures is not unique to China. Nvidia's AI Technology Center has explored similar ideas, such as developing AI-powered avatars of historical figures or loved ones, based on available recordings and images. Dr. Simon See, head of the center, has suggested the possibility of recreating historical eras or figures like Napoleon, given the availability of their letters and documented history. However, the notion of reviving controversial figures through AI has sparked debate, with concerns about the spread of their ideologies in modern discourse.
In China, strict internet censorship ensures that any unauthorized AI content, especially those critical of Xi, would be swiftly removed. The government's control over online content means that XiBot will likely serve as an extension of state propaganda, reinforcing Xi Jinping's philosophies while minimizing the risk of dissenting views emerging through AI platforms.
