The AI ‘Race’: China vs. the US with Jeffrey Ding and Karen Hao
Aug 31, 2023
auto_awesome
Jeffrey Ding, an academic researcher, and Karen Hao, a journalist, discuss the nuances of China's AI development, including advantages, limitations, and risks. They tackle topics such as China's surveillance and data collection edge, the relationship between Chinese companies and the government, concerns in AI development, challenges of coordination, and the complex AI competition between China and the US. They emphasize the importance of understanding China's progress and the potential for collaboration.
The US maintains an edge in overall AI development compared to China, due to factors like talent limitations, compute limitations, and chip shortages.
China lags behind in large language models (LLMs) and relies on the US for key chips required for AI development, making self-sufficiency challenging.
Deep dives
China's AI Development and the US-China Competition
China's development of AI is a topic of concern, with the US often fearing that China will surpass them. However, experts suggest that overall, the US is still ahead in AI development while China is behind. One area where China excels is in computer vision, thanks to extensive investment by the government and a data advantage from surveillance. But in terms of overall AI development, the US maintains an edge due to factors like talent limitations, compute limitations, and chip shortages. Additionally, the relationship between the Chinese government and the tech companies is complex, with some companies having a close relationship while others face friction. Overall, while there is competition between the US and China, the idea that China will overtake the US in AI is not supported by evidence.
China's Advancements in Large Language Models
China currently lags behind in large language models (LLMs), especially in comparison to the US. The Chinese language data is limited, with more high-quality scientific information available in English. Chinese researchers struggle to build highly performing chat GPT models purely based on data. Additionally, China still relies on the US for key chips required for AI development. While China is investing in semiconductor development, the complexity of the industry and global dependencies make it challenging to achieve full self-sufficiency. Overall, the US retains a significant advantage in frontier AI research and the development of LLMs.
China's Relationship Between Companies, Government, and Military
The relationship between Chinese tech companies, the government, and the military is complex. While some companies have close ties with the government, there is also friction and differing objectives. The Chinese government values information control and imposes strict regulations on AI technologies, particularly concerning censorship and data control. However, there are companies that push back against government demands to protect data privacy. It is important to recognize that Chinese companies and the government do not always align perfectly, and there is a range of relationships between them. Furthermore, companies in China often find innovative ways to maintain control while cooperating with government requirements.
The Importance of International Coordination on AI
The US and China's cooperation and coordination on AI is crucial to address the risks and challenges posed by this technology. International coordination, similar to regulating carbon emissions, could help ensure a safer and more sustainable development of AI. However, high-level political tensions and lack of trust make international agreement challenging. Nevertheless, track two initiatives, involving researchers, engineers, and industry leaders from both countries, show promise for facilitating coordination. These efforts allow for closer collaboration at the ground level, where technology builders can work together and find common ground on concerns like safety and security. International cooperation is seen as crucial to avoid an uncontrolled race in AI development.
In the debate over slowing down AI, we often hear the same argument against regulation. “What about China? We can’t let China get ahead.” To dig into the nuances of this argument, Tristan and Aza speak with academic researcher Jeffrey Ding and journalist Karen Hao, who take us through what’s really happening in Chinese AI development. They address China’s advantages and limitations, what risks are overblown, and what, in this multi-national competition, is at stake as we imagine the best possible future for everyone.
CORRECTION: Jeffrey Ding says the export controls on advanced chips that were established in October 2022 only apply to military end-users. The controls also impose a license requirement on the export of those advanced chips to any China-based end-user.
