Telecom firms strut latest LLM stuff
Immense data, rich scenario edges in country still require further exploitation
For example, Yuanjing LLM dramatically enhances design efficiency at a Hangzhou garment factory. Previously, a skilled designer could create three to four clothing designs per day, taking two days to finalize one. Now, the AI model generates designs in just three seconds, including visualizing the fit, vastly boosting designers' productivity, China Unicom said.
In home appliance manufacturing, Yuanjing LLM improves quality control through AI-augmented visual inspections, reducing defect rates by 50 percent. In port operations, China Unicom's AI applications, developed with a port company in Nanjing, Jiangsu province, enhance safety and efficiency across over 30 operational scenarios, significantly reducing accident rates.
China Mobile Chairman Yang Jie also highlighted the company's AI journey since 2013, culminating in the largest single intelligent computing center among global telecom operators, with a total computing power of 17 exaflops. An exaflop is a measure of performance for computers that can calculate at least 1 quintillion floating point operations per second.
Chen Yili, deputy chief engineer at the China Academy of Information and Communications Technology's Cloud Computing and Big Data Research Institute, said: "Algorithms, computing power and data are the three pillars that support AI development, with computing power serving as the driving force for AI advancement. As large-scale AI models now have 1 trillion parameters, demand for computing power currently exhibits exponential growth. The enhancement of computing power not only accelerates the training speed of large-scale AI models, but is also a crucial factor for the rapid iteration and widespread application of AI technology."
With over 10,000 AI plus projects across sectors like governance, industry and healthcare, China Mobile said its LLMs significantly contribute to digital transformation.
Moreover, China Mobile has established the Jiutian Research Institute, fostering a team of 2,000 AI experts. They aim to produce top-tier AI technological innovations, efficient industry integration platforms and high-quality AI+ applications to further empower the digital economy.
Xiang Ligang, director-general of the Information Consumption Alliance, a telecom industry association, said the initiatives of China's leading telecom operators illustrate a transformative approach to industrial modernization through AI and data integration.
As they continue to innovate and collaborate across various sectors, the potential for AI-driven efficiency and productivity gains appears boundless, representing a significant step toward a smarter, more automated future, Xiang said.
Ke Ruiwen, chairman of China Telecom, stated, "We will actively participate in the construction of the national data infrastructure, providing functions such as data confirmation and registration, data fusion computing and trusted data spaces, to support data aggregation, circulation and utilization."
This concerted effort by government bodies and enterprises marks a significant step toward realizing the full potential of data elements and AI, driving China's digital economy forward.