China's Generative AI Challenges Amidst Global Fragmentation
Saxo Bank 12.09.2023 11:10
China is confronted with challenges in the field of generative AI as it navigates a global order of fragmentation. The success of generative AI breakthroughs in the U.S., coupled with limited computing power and geopolitical tensions, has threatened to break down China’s virtuous cycle of technology application, productivity enhancement, and growth.
China's strategic initiatives for AI development
In May 2017, Google's AlphaGo astounded China when it defeated Ke Jie, the renowned Chinese Go champion and the world’s number one in ranking, with its artificial intelligence (AI) capability in a 3-0 victory. This triumph not only demonstrated the power of AI, but also somehow marked the point China embarked on a more aggressive pursuit of AI technology.
In July 2017, China unveiled its "Next Generation Artificial Intelligence Development Plan," followed by the "Three-year Plan for Next Generation AI Development (2018-2020)" in December. These strategic initiatives aimed to propel China to the forefront of AI innovation and were further complemented by the New Infrastructure Initiative introduced in May 2020.
China leveraged its mobile-plus-internet economy, fuelled by vast amounts of data, to empower AI algorithms in understanding consumer preferences. Internet giants like Baidu, Alibaba, Tencent and JD.COM successfully utilised AI to identify market trends, deliver personalised products and services, and even anticipate customer needs before they arise.
The challenge of generative AI
China has made significant strides in AI applications, but generative AI demands a new level of sophistication and computing power. Over the past two decades, Chinese companies thrilled on applying relatively mature technology to disrupt and transform how merchandise trading, retailing, payment, entertainment and social media are conducted, first starting by copying the business models from overseas and eventually coming up with business innovations that turned out to lead the global trend instead. Nonetheless, the achievement has been mainly about business model innovation and expanding AI applications to more and more aspects of people’s lives. In this process, the vast pool of data available in China has added to the speed and extent of development and made Chinese internet companies corporate giants with dominant domestic positions and global reach.
While China excelled in certain AI domains, particularly visual recognition and AI-enhanced big data analysis, it now faces a new challenge in the form of generative AI, similar to AlphaGo's breakthrough. Generative AI models, such as OpenAI's ChatGPT, have proven capable of generating coherent and contextually relevant content, expanding the boundaries of AI capabilities. This development has sent shockwaves across China, prompting a scramble to catch up.
In search of technology innovation and computing power
The Chinese authorities have reportedly been urging Chinese internet companies to come up with more generative AI solutions and applications and the latter are fully aware of the urgency to prepare for the disruption that generative AI presents. However, the new challenges posed by generative AI are the advancement of the underlying technology innovation and the massive demand for computing power, and less about business model innovation and new applications of mature technology. According to the Artificial Intelligence Index Report 2023 by Stanford University, China has come close to the US in AI conference citations (China: 22.02%; US: 23.86%) and depository citations (China: 20.98%; US:29.22%), but China (at 8.04%) is still a long way away from the US (at 54.02%) in terms of citations in the field of large language and multimodal models, which are fundamental to generative AI such as ChatGPT.
The challenge of computing power faced by China is even more daunting as computing power requires a large number of advanced microchips, which China is incapable of designing and manufacturing. Moreover, the US ban on exporting high-end semiconductors as well as the relevant equipment and technology to China since October last year has further exacerbated the situation, limiting the country's access to the necessary hardware for rapid AI development, in particular generative AI.