How China plays catch-up in the AI race with US
Technology specialist Yin Ruizhi assesses that the recent ban on Chinese companies’ access to OpenAI would not make much difference to the development of China’s AI large language model industry.
In July, AI large language model (LLM) leader OpenAI announced a complete ban on access to its interface in China, leading many media outlets to predict that the move would stunt China’s AI LLM development. In fact, China’s AI LLM industry has continued to develop rapidly even after the ban.
No signs of stagnation
We can look at the impact of the ban on China’s AI industry from two perspectives: small and medium-sized startups and large internet giants.
Small and medium-sized startups have quickly replaced foreign LLMs with domestic ones at low cost. After OpenAI announced the suspension of application programming interface (API) services in China, many domestic LLM companies, such as Zhipu AI, Tencent Cloud, Moonshot AI, Baidu AI Cloud and Alibaba Cloud immediately launched preferential migration services for OpenAI customers.
Several small startups in China have already switched to domestic LLMs, with many of them having a better understanding of Chinese language content than ChatGPT.
Currently, the APIs of top Chinese LLMs are not inferior to OpenAI in terms of everyday Chinese language tasks. Several small startups in China have already switched to domestic LLMs, with many of them having a better understanding of Chinese language content than ChatGPT.
The four primary fields for AI LLM entrepreneurship are education, customer service, office work, and nutrition and health. To effectively acquire users, the most crucial strategy is to deploy targeted advertisements across internet platforms.
The number of companies launching ads in these fields, along with the price and scale of the ads, are often used as key indicators by industry insiders to measure whether the wave of entrepreneurship in these fields is declining.
Before July, advertising costs for AI LLM office startup applications on internet platforms in China was around 1.4 RMB to 2.7 RMB (US$0.20 to US$0.38) per click, with over a thousand companies participating in the bidding every month.
In early August, the cost of a search for AI office applications was still around 2.7 RMB per click, with over a thousand companies participating in the bidding. Meanwhile, the cost of ads for AI-assisted powerpoint slides production has continued to soar to around 13 RMB per click.
The AI customer service field is even more popular, with the average cost of keywords reaching a high of around 20 RMB per click in August after OpenAI banned Chinese access, higher than the average of 16 RMB per click in July (to be fair, this price growth is also driven by the summer tourism boom).
The Paris Olympics became a significant event for Chinese internet technology companies such as Alibaba to showcase the charm of technology by using AI technology in broadcasts.
As China’s cultural tourism and service industries are starting to lower costs and increase efficiency on a large scale, ad launches in this field are still seeing a rise in both volume and price.
Clearly, the innovative applications by small and medium-sized enterprises in China’s AI LLM industry are not showing any signs of stagnating.
Olympic use-case
From the internet giants’ perspective, China’s AI industry has also shown no signs of slowing down. The Paris Olympics became a significant event for Chinese internet technology companies such as Alibaba to showcase the charm of technology by using AI technology in broadcasts.
On 31 July, Alibaba announced that its Olympic AI LLM Tongyi Qianwen, trained with professional knowledge about the Olympics to promptly answer related questions, would be publicly available for free. It is also equipped with the best capabilities in Chinese-French simultaneous translation — on par with professional interpreters — providing Chinese-French translations for on-site audiences at the Paris Olympics.
Alibaba Cloud also provided AI-enhanced multi-camera replay systems for 14 Olympic venues. The basic working principle was to collect images from multiple cameras, send them to the cloud, and use Alibaba Cloud’s AI computing power to perform real-time spatial reconstruction and 3D rendering. The selected moment would be replayed and generated in the cloud in just a few seconds, then seamlessly integrated into the live broadcast and shared around the world.
China’s unique investment and financing environment has caused it to lag behind the US in IT paradigm breakthroughs. But once the technological paradigm is clear, it is astonishing how quickly China’s IT industry can catch up.
International Olympic Committee president Thomas Bach praised Alibaba’s AI technology, saying that it has brought the Paris Olympics to new heights.
Implications on China-US tech competition
Indeed, the achievements of Chinese companies in the AI LLM field are often underestimated by traditional media.
When it comes to technology, people often confuse paradigm breakthroughs with application development. OpenAI led the paradigm breakthrough of AI LLMs, but this technology is now widely understood.
In the future, what will determine industry development is the development of technology in specific application fields. For example, cars were not invented by American companies, but before the Second World War, American companies such as Ford had become the core giants of the global automotive industry.
China’s unique investment and financing environment has caused it to lag behind the US in IT paradigm breakthroughs. But once the technological paradigm is clear, it is astonishing how quickly China’s IT industry can catch up. The competition between China and the US in AI technology is far from over.
