While workers across the Western world have been taking to ChatGPT and Bard in their millions since the former chatbot’s launch back in November 2022, developers in China have been hard at work playing catch up.
This week, it was revealed that China has given the green light to over 40 AI models to be used for public use over the last six months. The news followed an announcement that Baidu’s Ernie Bot would be incorporated into the latest Samsung Galaxy S24 model.
While a Chinese chatbot is yet to surface internationally in quite the same way as ChatGPT, these developments suggest that a capable rival might not be too far away.
China’s Chatbot Approval Spree
Chinese State-backed media organization The Securities Times has revealed that Beijing has granted approval for 14 large language models (LLMs) to be made available to the public during the past week.
This is the fourth time China has permitted companies to make their large language models available to the general population, with the first firms given the green light to do so back in August 2023.
Among the initial batch of Chinese businesses to be awarded the seal of approval were search engine Baidu, ecommerce platform Alibaba, and ByteDance, the company that owns TikTok and its Chinese counterpart Douyin.
This just in! View
the top business tech deals for 2024 👨💻
The AI Arms Race Goes Global
Yahoo News reports that China had 130 LLMs in development when ChatGPT was launched, “accounting for 40% of the global total and just behind the United States’ 50% share,” referencing data compiled by CLSA.
Ernie – an LLM recently incorporated into the Samsung Galaxy S24 – has over 100 million users, according to its creator Baidu, which is the dominant search engine used in China.
However, China has enforced stringent regulations on how AI chatbots can be used and developed within the country. Companies developing LLMs have to gain licenses from the central government to release their technology to the public.
AI in China: An Authoritarian Balancing Act
The rise of AI tools like ChatGPT presents a collection of difficult conundrums for governments that want to enforce tight restrictions over the content and information their citizens can see – such as China’s CCP.
China can’t afford to be left behind by the rest of the world when it comes to modern technologies it wants to lead it – but balancing this with a desire to control information in the way it historically has done is going to be tough.
Controlling the outputs generated by a large language model as powerful as Ernie or ChatGPT is a completely different ball game to implementing “The Great Firewall of China”, the country’s current web of internet censorship mechanisms that lets the government bans channels like HBO whenever it wants.
AI chatbots can “hallucinate” and produce downright bizarre responses drawn from information in their training data. Just like with ChatGPT, Bard, and Claude, there’s always the risk that they’ll say something you weren’t expecting.
They can be as unpredictable as human beings – and this is a big issue for China. Of course, LLMs can be constrained by altering the mechanics of their algorithms and knowledge bases – and, in China’s case, tight regulation and high levels of government oversight.
China is charting a fundamentally different course when it comes to AI development to many Western nations, and what is allowed to be produced – and used – under the close supervision of the ruling regime is likely to have significant political ramifications inside and outside of the country’s borders.
