China creates LLM trained to discuss Xi Jinping’s thoughts

🌈 Abstract

The article discusses the development of a large language model and conversational AI system called "XiBot" by China's Cyberspace Research Institute, which is based on the philosophies of President Xi Jinping. The article also touches on the broader implications of using generative AI to recreate historical figures and the potential risks and concerns around such technology.

🙋 Q&A

[01] China's XiBot

1. What is the XiBot?

• The XiBot is a large language model and conversational AI system developed by China's Cyberspace Research Institute.
• It is based on the philosophies of President Xi Jinping, with a corpus of government documents and the book "Xi Jinping Thought on Socialism with Chinese Characteristics for a New Era".
• The XiBot is currently being tested internally and is not yet available to the public.

2. What is the purpose of the XiBot?

• The XiBot is designed to allow citizens to query government documents using a generative AI interface.
• It is an extension of the Chinese government's efforts to ensure that Xi Jinping's thoughts and philosophies are widely known both domestically and internationally.

3. How does the XiBot differ from generative AI systems in other countries?

• In China, where Xi Jinping's thoughts are considered infallible, the risk of AI hallucination and the spread of misinformation is potentially lower than in other countries.
• The article suggests that any unauthorized AI system in China would be quickly shut down and the creators would face unpleasant punishments.

[02] Recreating Historical Figures with Generative AI

1. What are the potential implications of using generative AI to recreate historical figures?

• Researchers have discussed the possibility of using generative AI to recreate historical figures, such as Roman times or Napoleon, based on available records and documentation.
• However, the article raises concerns that some may welcome the recreation of historical tyrants, and the spread of their ideologies through social media could be problematic.

2. How does the availability of video footage impact the potential to recreate contemporary historical figures?

• The article suggests that contemporary historical figures are better documented than historical figures from earlier eras, due to the availability of video footage.
• This means the potential to recreate them as conversational AIs is greater than for figures from earlier periods.

3. What are the family's views on the idea of recreating historical figures using generative AI?

• The author's family declared the idea "daft", arguing that it could lead to the spread of the vile ideologies of historical tyrants, which would be undesirable.