New Chinese chatbot was trained on President Xi Jinping's philosophies on socialism and several other 'major internet information' databases

Chinese President Xi Jinpeng
(Image credit: Shutterstock)

China is testing a new large language model (LLM) and conversational AI that will, among other things, allow the country’s citizens to ask President Xi Jinping how to be better socialists. The country’s Cyberspace Research Institute recently announced the chatbot is being tested internally.

The state agency organized a team of experts to develop the chatbot, which was trained using the data from its Cyberspace Research Database. But it’s not just a “What would President Xi do” model. The corpus of information used to train the model also includes “seven major Internet information professional knowledge bases.”

The LLM was built using a “registered domestic open source commercial pre-trained language model,” so it’s not OpenAI’s GPT, Google Gemini, or any of the other well-known AI chatbots. Even once it's released from internal testing, the LLM will be deployed via a dedicated server within the China Cyberspace Research Institute.

China’s Cyberspace Research Institute incorporated President Xi’s book, Xi Jinping Thought on Socialism with Chinese Characteristics for a New Era, which is recognized as the president’s definitive political philosophy. Other listed works included in the LLM are the "Network Information General Knowledge Base," "Network Information Dynamic Knowledge Base," "Network Information Journal Knowledge Base," and "Network Information Report Knowledge Base."

China Cyberspace Research Institute says users can choose which knowledge base category to use, and can engage the chatbot in an intelligent question-and-answer session. The AI chatbot can also generate manuscripts, summarize bodies of text, and perform translation between Chinese and English, among other tasks.

The Register pointed out an interview with Dr. Simon Lee, global head of Nvidia’s AI Technology Center that the idea of deliberately incorporating a specific figure's ideas and philosophies isn't new. In that interview, Dr. Lee admitted that Nvidia hopes to one day collaborate with researchers to create generative AI-powered replicas of other important historical figures, such as Roman emperors or even Napoleon.

Jeff Butts
Contributing Writer

Jeff Butts has been covering tech news for more than a decade, and his IT experience predates the internet. Yes, he remembers when 9600 baud was “fast.” He especially enjoys covering DIY and Maker topics, along with anything on the bleeding edge of technology.

  • Geef
    Hopefully he doesn't start relying on a AI bluescreen for his speeches. There are some funny videos of people zoning out and just reading what is on the screen without thinking first.

    Xi Jinping voice: "... and I believe this country is great... end sentence next paragraph."
    Reply
  • MergleBergle
    "What is the problem with a multi-party democratic system?" "DOES NOT COMPUTE! DOES NOT COMPUTE! ATTACK, ATTACK, ATTACK! EVEN THOUGH TAIWAN HAS NEVER BEEN PART OF THE PRC, IT BELONGS TO THE PRC!! BECAUSE PEOPLE OF CHINESE ANCESTRY LIVE THERE! EXTERMINATE! EXTERMINATE!
    Reply
  • Notton
    This is funny, but kind of interesting at the same time.
    It's like those heads in jars from futurama.

    US should make one for Lincoln and Roosevelt
    Reply
  • Pierce2623
    You literal couldn’t make this stuff up.
    Reply
  • bit_user
    Xi Jinping can't be everywhere and doesn't have time for everyone. However, through the miracle of AI, each citizen can get personalized attention from their own Big Brother.
    Reply
  • pcbulk
    Communist philosophy from a defacto dictator.
    Reply