Last week, Shanghai hosted China's largest AI event: the World Artificial Intelligence Conference (WAIC), with 500 exhibitors, 1,500 exhibits, more than 300,000 attendees, and even Chinese Premier Li Keqiang in attendance. of
But despite its impressive scale, the conference left me disappointed. I look forward to witnessing the technological advancement of this sector. Instead, WAIC confirmed my suspicions: there is a gap between what China's AI can do and the cutting-edge innovations emerging from Silicon Valley.
WAIC exhibitors focused on robotics and large language models (LLMs), with only a few creative AI companies participating. More than half of the companies at WAIC, including major tech giants and even some state-owned telecommunications companies, were exhibiting their new models.
In Shanghai, Baidu founder Robin Li encouraged participants to start developing practical AI applications instead of refining their LLMs. A powerful and widely used AI application will benefit society more than another model that can process vast amounts of data but has no practical use, he asserted.
The AI applications on display in Shanghai were mostly chatbots like ChatGPT, except Kwaishu's text-to-visual application Killing, a Sora-like product that I found really impressive.
As I was walking around the showroom, I noticed that most of the chatbots required a prompt in English instead of Chinese. This leads me to suspect that many of China's AI programs are actually running on models developed outside of China.
It's clear that the models still need some fine-tuning. One user pointed to a text-to-visual app from MoreThreads that features “a cute baby with brown hair sitting in a garden.” The result was a baby with shiny skin, eyes that didn't line up on the face, and a disproportionately small body.
I left the conference earlier this year agreeing with Alibaba chairman Joe Tsai's frank admission that China's creative AI development is at least two years behind the US, meaning that American and Chinese companies are truly one. Not playing in the same league, and therefore difficult to do directly. Compare them.
The main problem is that China's LLMs are limited to using data inside the Great Firewall. As investment bank Goldman Sachs noted late last year, “LLM performance improves with scale—more parameters, more refined training data, more training runs and more computations.” An isolated Chinese-language Internet has much less information than an open Internet with sources in many different languages.
AI companies outside of China still have a lot of data they can use for training. An AI developer in China will struggle to keep pace.
The constraints posed by limited access to high-end GPUs are also clearly evident. US policies that reduce access to the latest chips and chipmaking technology will mean that Chinese companies lag behind their non-Chinese peers.
Yet despite these limitations, China's AI developers are finding opportunities for innovation.
A strong talent pool is moving into AI from the country's mature consumer tech ecosystem. Most of the founding members of the hyped “Four Tigers”—Baichuan, Zippo AI, Moonshot AI and Minimax—had worked at a large tech company. Their strong insight into consumers and products is why they now lead China's AI application space. From a consumer perspective, their products are on par with many of America's leading applications.
There is progress on the hardware front as well. Huawei's Ascend AI processors, in particular, appear to be miles ahead of their competitors. The Chinese tech company, now using chips developed by SMIC, claims that its Ascend 910B AI chip can outperform Nvidia's A100 chip in some tests, especially when used to train large AI models. .
Chinese AI developers face some fundamental obstacles, such as a challenging environment, lack of advanced chips, geopolitical isolation, and national security concerns that limit the movement of talent and capital.
Together, these constraints will create two parallel AI ecosystems: one inside China and one outside it. The United States will maintain its lead in developing this transformative technology.
But just because the US has the technological lead doesn't mean China's AI developers will be left behind. Chinese companies have always started out a step behind their non-Chinese peers, yet fierce competition and a willingness to experiment have helped them gain global reach and, in the case of consumer Internet companies, even the competition. me also.
In the world of AI, the US and China are both enemies and rivals. We should hope that geopolitical competition between them does not hinder innovation and cooperation.
Opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.