This article is more than 1 year old
China's tech giants and Beijing – the city – rush to build AI chatbots
Bootleg ChatGPT mini-apps are already testing the limits of OpenAI's policies
Beijing Municipal Bureau of Economy and Information Technology said on Monday it will support enterprises in building large AI models that compare to ChatGPT, as China's tech giants rush to deliver their own generative AI chatbots.
The agency announced it would support enterprises to build the models as well as an ecology of open source frameworks.
The bureau also wants to strengthen AI infrastructure, accelerate the supply of foundational data, and guide the collaboration of enterprises, universities, research institutions and open source communities.
The agency reckons Beijing's the right place to build an indigenous ChatGPT – it is home to 29 percent (1,048) of China's AI companies.
The Chinese government, well known for harsh regulations on Big Tech over the past two years, said last August it would support "the healthy development of internet enterprises."
- China's Baidu reveals generative AI chatbot based on language model bigger than GPT-3
- China cybersecurity regulator wants to support tech growth
- China's internet regulator details algorithms used by local Big Tech players
- You get the internet you deserve
Unlike many western web sites – including Google, Facebook, Twitter and Wikipedia – The OpenAI chatbot, ChatGPT, is not blocked in the Middle Kingdom. It is however, difficult for Chinese netizens to get an account on the service as users can't sign up with Chinese phone numbers. Thus its usage falls into a murky gray area legally.
Companies like Tencent and its WeChat have played whack-a-mole to keep mini programs based on ChatGPT off of their platforms. Searches on WeChat for ChatGPT are currently fruitless.
ChatGPT can generate text in over 90 natural languages, including Chinese, and some programming ones too. It also can translate between natural languages. Its performance, however, will vary among them, as the app is designed primarily for English. Beyond just using it to play around, firms want the technology integrated into their own products.
Therefore, it makes a lot of sense for China to want an indigenous version of the app – particularly since the central government will certainly want to manage and censor its output. Surely the CCP doesn't want to know what the training material has to say about Xi Jingping.
Chinese firms have scrambled to capitalize on the excitement and popularity of ChatGPT. Earlier this month, the country's web giant Baidu revealed it will launch a generative AI chatbot later this year. And it's certainly not the only one.
Alibaba said its research institute, Damo Academy, is conducting internal testing on an AI chatbot. No timeline was provided for the product.
Tencent Holdings, NetEase and JD.com also have their own equivalents in development.
And then there are the less salubrious or more obscure platforms that are emerging – some even ripping off ChatGPT's name and/or branding.
Peking University sociology professor Jiang Ruxiang attempted to explain why China lagged behind on AI language development. According to a translation, it all falls back on the structure of Chinese science and tech startups.
Jiang said tech orgs are motivated by rapidly becoming profitable, and that it's not clear how locally-developed natural language processing tech will deliver that – especially when the tech can be imported and modified. ®