Is China still Relying on US AI Tech?

According to a recent report from Reuters, the Biden administration is planning to shield US technology from China and Russia. To be more specific, the Commerce Department will focus on limiting the export of advanced AI technologies. These technologies will include OpenAI’s ChatGPT and closed-source AI systems trained on undisclosed data.

The real question is, how advanced are Chinese AI technologies at the moment? And are they relying on US tech to build their domestic AI models? Let’s find out.

How advanced is Chinese AI tech? 

When OpenAI released ChatGPT in late 2022, everyone was surprised, including China. In 2023, Chinese companies began releasing AI generative chatbots after getting approval from their government. OpenAI is unavailable in mainland China, so how did this happen? 

Many Chinese companies and engineers have gained access to ChatGPT through proxy servers and VPNs. From there, they managed to build their own applications and software on top of OpenAI’s models. Most Chinese companies benchmark their models against OpenAI’s models.

Last February, OpenAI disrupted two China-affiliated threat actors known as Charcoal Typhoon and Salmon Typhoon from accessing ChatGPT. These companies used OpenAI’s services to translate technical papers, debug code and generate scripts. Last December, OpenAI also suspended accounts belonging to the Chinese giant ByteDance because they used ChatGPT to create a rival AI model.  

China is playing catch-up in the artificial intelligence sector. There’s a big gap between OpenAI and Chinese AI technology. Some Chinese companies have achieved success in building AI models, but not like ChatGPT. The reason is that China lacks access to data since most available data sources are in the English language.  

We disrupted five state-affiliated malicious cyber actors’ use of our platform.

Work done in collaboration with Microsoft Threat Intelligence Center. https://t.co/xpEeQDYjrQ

— OpenAI (@OpenAI) February 14, 2024

How would China build AI models?

China is facing difficulties creating AI models since there’s a restriction on accessing closed-source AI models and advanced GPUs. In 2022, Nvidia and AMD banned China from purchasing their most powerful AI chips. The ban on software and hardware access did not stop the East Asian country. 

Regarding software, China is relying on open-source AI models like Meta’s Llama LLM models. According to Wired, a new startup named 01.AI achieved massive success after releasing Yi-34B on HuggingFaces. Yi-34B surpassed Llama 2’s scores and became one of the recommended models for developers. However, 01.AI received backlash after developers discovered that Yi-34B is based on Llama. 

Regarding hardware, Huawei is leading a group of semiconductor companies in manufacturing local AI chips. The Chinese company is competing with Nvidia and trying to build advanced AI chips autonomously. It’s working on designing and manufacturing high-bandwidth memory chips from the ground up, which are critical parts of advanced AI GPUs.

China is still behind in the AI race, and has a long road ahead. OpenAI has a head start because creating an AI model involves training algorithms and accessing a large data pool. The East Asian country lacks in both areas and will continue to depend on Western technology, at least for now.