ChatGPT is woke, and the Chinese communist is trained on the thoughts of Xi Jinping

It is well known that China likes to control its population, but what they have done now is truly insane. ChatGPT is banned in China, but that does not prevent the Chinese authorities from working on their own AI chatbots that will only say what Beijing wants. And to ensure that nothing inappropriate comes out of the ‘mouths’ of the chatbots, it has been decided to train them exclusively based on the thoughts of Chinese President Xi Jinping.

The latest large language model in the country has learned from the political philosophy of its leader, known as ‘Xi Jinping Thought on Socialism with Chinese Characteristics for a New Era’, as well as from other official literature provided by the Chinese Cyberspace Administration (CAC).

– The expertise and authority of the corpus ensure the professionalism of the generated content – said the Chinese Cyberspace Administration.

Although it has not yet been released for wider use and is only used in a research center, it is only a matter of time before CHATXi GPT, or whatever it is called, sees the light of day. The new model can answer questions, create reports, summarize information, and translate between Chinese and English, according to the announcement.

The creation of the new large language model followed extensive efforts by Chinese officials to spread Xi’s ideas on politics, economics, and culture in various formats. More than a dozen books have been published in Xi’s name, and his bestsellers usually take center stage at book fairs in the country. Popular news apps from companies like Tencent or NetEase reserve spots at the top of user feeds for articles from official media, where Xi appears most of the time.

Officials have also required school children as young as 10 to learn his political philosophy. They created the Study Xi Strong Nation app to teach and test knowledge for about 100 million party members in the country. In 2018, his ideas were also inscribed in the state constitution.

Attempts to ensure that artificial intelligence understands Xi’s philosophy come as Chinese officials try to balance draconian controls on freedom of speech in the country with encouraging the development of artificial intelligence.

AI Socialist Values

CAC, which has been at the forefront of issuing rules for generative artificial intelligence and introduced a licensing regime, mandates that generative AI providers ’embody core socialist values’ and states that generated content cannot ‘contain content that undermines state power’.

This poses a particular challenge for model developers due to the relatively scarce datasets available for training their large language models in the Chinese language. Most groups have trained their models on English-language information, introducing the potential for generative artificial intelligence to produce responses that contradict Chinese speech norms.

Tech giants like Baidu and Alibaba have ensured that their models strictly control generated content related to Xi or other potentially sensitive issues. Generative AI chatbots from both groups typically ask users to restart chats when pressed with sensitive topics.

To help developers cope with this issue, the Chinese Cybersecurity Association, a non-profit organization affiliated with the CAC, has released the first public database of 100 million entries of ‘high-quality and reliable data’ for groups to use in training models. The training set heavily relies on government regulations and political documents, reports from state media, and other official publications, according to sections reviewed by the Financial Times.

One of the dozens of textual documents in the data package contains 86,314 mentions of Xi Jinping.

– We must ‘ensure that in thought, politics, and action, we are always in high compliance with the Central Committee of the Party with General Secretary Xi Jinping at the center’ – concludes the publication.