Meng Pu, chairman of Qualcomm China, said that currently, the interaction and integration between 5G and AI have become a major trend in technological development. Meng Pu believes that with the support of 5G, with the rapid popularity of generative AI and the increasing computing needs, the importance of hybrid processing has never been more important.​

Meng Pu said that unlike processing only in the cloud, hybrid AI distributes and collaboratively processes AI workloads between the cloud and edge terminals, so that AI can be scaled up and realize its huge potential. The hybrid AI architecture can choose different ways to allocate task load between the cloud and the terminal side based on factors such as the complexity of the model and query requirements, or it can be completely performed on the terminal side; for more complex tasks, the model can be spread across the cloud and run it on the terminal. Hybrid AI will deliver cost, energy, performance, privacy, security and personalization advantages on a global scale. The cloud and smartphones, cars, personal computers and IoT terminals work together to achieve more powerful, efficient and highly optimized AI.​

When AI processing capabilities are extended to the edge, context-rich data is generated on the edge, and the intelligent connected edge comes into being. This will promote the large-scale deployment of richer AI applications and improve cloud intelligence overall.

"Terminal-side AI is the key to realizing hybrid AI architecture and expanding generative AI to a wider range of the world." Meng Pu said that compared with running AI on the cloud, terminal-side AI has the advantages of privacy, reliability, low latency, efficiency and personalization. There are many advantages in terms of culture. (Xinhuanet)