China is positioning itself to become a global leader in AI large models, despite initial setbacks. Experts believe the country has the potential to surpass the United States in this field and compete for global dominance.
At the 2024 Large Model Technology and Application Innovation Forum in Beijing, Zheng Weimin, an academician at the Chinese Academy of Engineering, highlighted two key trends in AI development: foundational large models evolving into multimodal models that integrate text, images, and video, and the rise of “large models+” that are being applied across industries like finance, healthcare, automotive, and manufacturing.
Zheng emphasized the importance of developing a domestically produced, high-performance computing system with AI acceleration chips like GPUs and TPUs to support large model systems. He outlined the lifecycle of large model development, which includes data acquisition, preprocessing, model training, fine-tuning, and inference. The foundational models are customized for specific industries, like healthcare, by fine-tuning them with specialized data, such as hospital records or ultrasound data.
Yu Youping, president of Beijing ZKJ Technology Co., discussed the shift from rapid advancements in AI to a focus on practical, enterprise-level applications. He stressed that the industry is now moving toward fine-tuned implementations that provide real solutions to business problems, where the combination of platforms, applications, and services plays a key role in implementing large models effectively.
NEWS DESK
PRESS UPDATE