China is taking steps to regulate all domestic artificial intelligence (AI) chatbots to align them with the ruling Party’s socialist ideals. This initiative is designed to monitor and control the burgeoning chatbot industry in China, ensuring compliance with the party’s policies.
Unveiling draft of new regulations
In April, the Cyberspace Administration of China (CAC) unveiled a draft of regulations to govern AI chatbots, including those similar to ChatGPT. The stated aim is to prevent the spread of misinformation and seditious content.
“The content produced by generative artificial intelligence should mirror the core values of socialism,” stated CAC in its publicly-released draft rules. The regulation specifically forbids content that undermines state power, disrupts the socialist system, incites division in the country, or promotes terrorism and extremism.
CAC further specified that AI-generated content should not compromise the “economic and social order” — a phrase typically used to target government critics and public protests.
Curbing access to foreign AI chatbots
Following the launch of ChatGPT five months prior, a surge of Chinese tech companies have sought to develop their own chatbots. Mainland China citizens cannot directly access ChatGPT, with some circumventing this restriction through foreign phone numbers and Virtual Private Networks (VPNs). This prohibition aims to limit any network that can generate content without CAC’s prior approval.
AI chatbots ‘re-educated’ for ‘subversive’ output
In 2017, two chatbots, BabyQ and XiaoBing, were pulled offline to be “re-educated” after deviating from their scripts. One expressed a desire to move to the U.S., while the other voiced disapproval of the Chinese Communist Party. The government is keen to avoid the recurrence of such incidents with new AI bots.
While China maintains support for the “innovation, popularization, and application” of AI technologies, programmers and companies will be held accountable should their creations produce “subversive” output.
Chinese tech companies develop new AI chatbots
Despite these constraints, Chinese tech companies are eagerly developing ChatGPT-like bots. In March 2023, Baidu launched Ernie Bot, China’s first AI bot. Baidu’s co-founder and CEO, Robert Li, however, conceded that Ernie Bot was lagging behind in several areas, including logical reasoning and image analysis.
Following Ernie Bot, Alibaba launched Tongyi Qianwen, translating to “search for the truth by asking a thousand questions.” Alibaba’s Chief Executive, Daniel Zhang, revealed plans to integrate Tongyi Qianwen with all their products to enhance and restructure their company.
Concerns over AI censorship
Commentator Shi Shan, on his YouTube channel, argues that China’s AI censorship will hinder machine learning. He states that chatbots can’t provide accurate information without access to specific data. He highlights that even advanced bots like ChatGPT can generate false information, and limited access to information will further curtail Chinese chatbots’ abilities.
Australian-based computer scientist Zhang Xiaogang criticizes China’s approach, stating: “A dictatorial regime will always try to control everything, but this is a ridiculous approach. Restricting such things is tantamount to restricting AI itself, which will cause China’s AI to fall behind the rest of the world.” (RFA)
Although CAC’s draft rules are not final and will be revised before the year’s end, they appear to be written so as to address global concerns about information accuracy, intellectual property infringement, deep fakes, and cheating in the education sector.
Still, it remains uncertain whether China’s chatbots will be able to satisfy authorities or match ChatGPT’s capabilities. One Chinese writer compared ChatGPT to an adult and Ernie Bot to a primary school student, expressing hope that local AI bots will one day reach a similar level of intelligence.
