LLMs in China and how to deal with China's AI strategy?
A few days ago, US President Joe Biden and Chinese President Xi Jinping met in San Francisco - on the doorstep of Silicon Valley. In addition to the many international crises, the topic of AI technology certainly played a role. The USA wants to slow down the Middle Kingdom. And China is committed to its own AI strategy, including LLMs.15 Nov 2023
In China, the results of LLM, generative artificial intelligence, must be in line with the socialist goals of the Communist Party. Only socialist content or content that does not contradict socialist ideas is permitted as results. Disbelieving looks on the faces of many managers during an internal event of a large mechanical engineering company when this fact appears on the screen. Is this a joke? No. China has its own AI (artificial intelligence) strategy. Recommender algorithms that suggest content to users in social networks, on websites and in chats must also meet the Communist Party's requirements. According to many analyses, China has quickly discovered the opportunities offered by AI - in the public service sector, but also in industry in the form of higher productivity. "But the government also quickly realized that there are algorithms that can mobilize people," explains Rebecca Arcesatis from the Mercator Institute for China Studies in the Industrial AI Podcast. They want to prevent this. The Chinese government has a say when internet platforms develop and deploy algorithms. China expert Arcesatis and her team are in great demand. The industry and politicians need help.
In contrast to many European countries, China prioritizes the topic of artificial intelligence high up in its political and social plans - closely followed by open source and robotics. The state invests billions, companies receive grants if they conduct research in the field of AI and develop products that solve real industrial problems, researchers write one AI paper after another and universities and companies apply for a large number of patents. Some Western observers speak of an aggressive patent strategy. "They are trying to patent everything in the AI sector, clogging up the patent offices and pursuing a very aggressive strategy," explains one industry manager. This is not illegal, but many smaller, innovative European companies in particular see this as a problem. They complain that Europe not only lacks an AI strategy, but also a patent strategy. Back to China: "We are experiencing an AI wave with many new companies," explains Arcesatis. In 2022, Chinese companies raised more private money in the field of AI than companies from Europe or the UK. The USA leads in terms of investment, but China is still catching up. However, China's AI strategy is also very political. The technology is helping industry, making the Middle Kingdom even more competitive, but the military is also an important player. AI is a key technology for the modernization of the armed forces. AI therefore also plays an important role in China's geopolitical ambitions - also thanks to European research?
Many people in Europe and especially in the USA are therefore worried. Relations between the USA and China are close in the field of AI research. More than 100 labs in Europe are also working together with Chinese researchers. This involves industrial AI applications, basic research, AI theory and natural language processing. There is also a great deal of cooperation in scientific publications. After the USA, the United Kingdom is the largest partner of Chinese scientists. However, European and US universities are concerned about cooperation. It is a matter of national security and the question of whether human rights in China are being violated by research from Europe or the USA. Where do the results go, who uses them and how? Politicians are called upon to create new rules. And companies are also asking themselves the question: are IT and computer science apolitical?
According to analysts, there has been a decline in scientific co-publications by US and Chinese researchers, but this could also be due to COVID-19. However, the tensions between the USA and China are real. Some Chinese researchers have left the USA, reports Arcesatis. People in Beijing are happy about this. Digital sovereignty was already at the top of the political agenda in the early 2000s. The government saw the danger of dependence on the USA in the IT sector. A separate open source ecosystem was created, and the Android ban proved the Chinese right. Today, no GPU unit crosses Chinese borders, urgently needed ASML machines of the latest generation are in European ports and may not be exported. The US government stopped the export of the important AI infrastructure. Chips are China's Achilles heel in the field of AI. Researchers are working on alternatives, combining chips, developing their own or working on AI applications that manage with less computing power - a GPU bypass, so to speak. Research into LLMs is also picking up speed in China - provided the results are content that is socialistically acceptable. Can this work? Back in the 1990s, Western observers predicted that China would never succeed in becoming a digital nation without a free Internet. They were very wrong. What does Europe need to learn? Companies and researchers need to better understand the Chinese innovation ecosystem. How do laboratories, researchers, companies, the party and security authorities work together, where are the connections, how is the ecosystem developing and where are the dangers lurking for researchers and entrepreneurs? You can't ask this of a research assistant at a university. Politicians are called upon to provide guidelines and clarification. The Chinese know our system very well, says Arcesatis. Europe still knows too little about China.
Interested in news about exhibitors, top offers and trends in the industry?
Your web browser is outdated. Update your browser for more security, speed and optimal presentation of this page.Update Browser