China’s 5G edge to boost AI
Country has an advantage in network speed, data volume
China’s early planning of 5G network, which will be 10 to 20 times faster than the existing 4G network, as well as the data generated by this infrastructure will give the country a head start in developing artificial intelligence (AI), company executives and analysts said on Thursday.
“The development of 5G standards is closely related to that of AI because the two technologies complement each other. The faster connection speed will generate more data, based on which AI industry could grow faster and function more efficiently,” Sun Gang, global vice president of US-based Qualcomm, said at an AI forum in Beijing on Thursday. He estimated that the global market volume of 5G-related products and services would reach $12.3 trillion by 2030, and the market size for AI-derived industries could hit $5.1 trillion by 2025.
As such, China could have an overwhelming edge in developing AI in the 5G era, as one of China’s advantages in the AI industry now lies in its huge data, and such data is likely to scale up and improve in quality with 5G infrastructure, industry insiders said.
For example, Sun Li, vice president of Chinese smart platform provider Thundersoft Technology Co, told the Global Times on the sidelines of the forum that China’s early start in the 5G era bodes well for the country’s development of autonomous driving, a sector that needs real-time decision-making to reduce driving risks. Thundersoft is the organizer of the forum.
“The highly efficient network will also buy some time for the AI algorithm [in the cloud] to analyze interactions between vehicles and vehicles, vehicles and roads, and vehicles and buildings. This can provide a much more complete and accurate roadmap for vehicles,” Sun added.
Work with embedded AI
China’s 5G plans would also help embedded AI “collaborate more harmoniously” with cloud AI, industry insiders said.
This year has witnessed a surge in embedded AI, a technology that can process data and run AI algorithms on devices without transferring data to cloud servers. The technology is now seeing wider application in such sectors as smart home appliances and security cameras, which require real-time transmission, processing and analysis, according to Sun.
“Developing hardware, like batteries and AI chips, as well as software to support in-device data compression and analysis and perform more functions is much more technologically challenging than developing a regular gadget, and it is also sometimes way too expensive for us to afford,” Sun said.
If the 5G network is fast enough, developers could move some analyzing procedures to the cloud AI server to ease the pressure. “We can also weigh the costs and benefits to decide which parts should be conducted within the device, and which parts to be processed in the cloud server,” Sun noted.
Data protection
But a pressing issue associated with the AI industry under the faster 5G network is security.
“Every driverless car is connected to the Internet. So what if someone hacks and controls the central system? The consequence could be catastrophic,” Jiang Su, a professor at Peking University, said at another forum co-hosted by Peking University and Stanford University in early December.
Si Xiao, head of Tencent Research Center, also highlighted the problem of protecting personal privacy when developing AI technology.
“We need to provide personal data when training AI algorithms. So it is necessary to design it into a private system to avoid the abuse of
data,” Si noted.