Several days before the double eleven shopping carnival in 2015, the first time I went to Alibaba‘s Tmall headquarters in China’s city of Hangzhou, one of the marketing VPs showed us real-time images of Taobao transactions on a smart screen. The giant e-commerce platform that Jack Ma built unexpectedly struck us with its high efficiency and volume, however few contemplate the real industrial revolutions it enables. This year at the Aspara Conference, Taobao and Tmall’s presence were limited. Instead the conference was wildly informative and technically focused on AliCloud, the infrastructure and operating system that fuels the innovations in various industries.
At the Aspara Conference in 2016, Jack Ma initiated the idea of “five new” strategies including new retail, new manufacturing, new finance, new technology and new energy. Under his guidance, the “five new” strategies today have evolved into innovations in numerous industries. “Up until today, the five new has generated hundreds and thousands of new innovations. The commonality behind it is the digitalization and intelligence of all industries at all dimensions.” Said Daniel Zhang, Executive Chairman and Chief Executive Officer of Alibaba and successor to Jack Ma. Over 85 new technological products were presented on the website of this year’s conference.
Alibaba Cloud — Building the infrastructure for industrial partners
This year marked the fifth year of the Aspara Conference, which also marks the 10 year anniversary of the Alibaba Developers’ Conference. Ten years ago in 2009, as pointed out by Dr. Wang Jian, one of the chief engineers of Alibaba and founder of Alibaba Cloud, the Aspara Conference used to be merely a local website development forum attended by website managers around the country. “Looking back, we could tell that the development process of the Aspara Conference echoed with the development of Chinese digital economy. Back then, we were still talking about the budding of cloud computing and big data.” Daniel Zhang commented.
However with the prosperity of data and intelligence, responsibilities fall onto the cloud computing platform of Alibaba. As Daniel Zhang mentioned in his keynote speech, “There are many cloud computing companies in the industry, as well as several large cloud computing participants and constructors around the world, but today Alibaba’s cloud computing is not only the IT infrastructure. We take into consideration the integration of cloud and edge computing, and aim to build a highly interactive intelligent network that combines cloud and edge computing.”
From the very beginning, Alibaba is much more than just an e-commerce company. The company has been building the IT infrastructure for 16 years now. Up until now, it has opened its cloud computing services to 19 regions and 59 data centers around the globe. Put simply, the clients and partners of Alibaba are able to share the mighty computing power together with the scientists of DAMO Academy, the tech giant’s thinktank. Together, the Apsara operating system and infrastructure makes the Alibaba cloud today, aiming to help small and medium sized businesses enjoy the fruits of Alibaba’s technological achievements.
Aspara, a massive-scale cloud computing operating system with the capability of clustering thousands of PCs, took its name from traditional Chinese mythology, as humans anticipated ascending into outer space and the great unknown. Another product that also took its name from Chinese culture is the new AI chip Hanguang 800.
AI chip Hanguang 800
According to Qi Xiaoning, the Vice President of Alibaba Group and one of the chief scientists, “Chips are the foundation of AI computing. Chips are like the brain, the heart, the infrastructure. Algorithm is the soul. AI development is dependent on the combination of the body and soul.” The Hanguang 800 chip is powerful in its incomparable image processing power of 78000 per second, with massive computing power equivalent to ten traditional GPUs.
According to Qi, the chip enables optimization for computer vision applications including classification, object detection and segmentation. For instance, on the Taobao app, there is an option named “photograph and search” that allows users to take photos of whatever product they see and search for similar items in the app. There are altogether 1 billion new photos added into the image gallery each day. The recognition process for traditional GPU took as long as one hour, meanwhile, it takes Hanguang 800 only five minutes to complete the task.
Industrial applications of Alibaba’s AI innovations
The applications of AI vary from the municipal government level to enterprises and ordinary consumers. Among them, one of the most talked about has been the city brain, first implemented around three years ago in Hangzhou. According to Zeng Zhenyu, VP of Alibaba Cloud Intelligence and expert on Alibaba’s industrial brain and city brain, “The city brain is built on top of the Alibaba Aspara Operating System, which provides a city-level data middle station. The middle station, is a word invented by Alibaba. It’s like a strong foundation for a city, which helps the government integrate and combine the data from all walks of urban life.” A most practical example would be Alibaba’s collaborations with Hangzhou Xiaoshan International Airport. Radar data, flight broadcasting data together with image materials keep the airport staff aware of the exact location and status quo of the planes. It would therefore save time for travelers by elevating the overall rate of on-time arrivals.
“Today, over half of China’s 20 most populous cities have installed urban brains. Our goal is that, in the upcoming three years, over 50% of China’s urban population can enjoy added convenience from the city brain.” Zeng said.
AI also helps elevate the efficiency in industrial quality inspection. Industrial component defects come in varied forms , allowing AI to easily outperform humans. According to Zeng, Alibaba worked with Zhengtai New Energy in the quality inspection of solar cells. The accuracy of human quality inspection is 93% to 96%, while algorithm achieves a rate of 98%, with double the speed. The same goes for medical services. AI works better with the segmentation and extraction of diseased organs, which requires higher accuracy when it comes to saving lives.
In terms of logistics, one example that has already been extensively implemented is terminal logistics, referring to the last one to three kilometers of delivery. Unmanned delivery vehicles are now applied in many Chinese universities. A student would receive a phone call from the unmanned vehicle, being informed of the exact time it arrives at the dormitory. Cainiao, Alibaba’s logistics arm, is also utilizing AI to optimize the delivery routes and even the packaging methods. As a result, during this year’s double eleven shopping carnival, less materials would be wasted from the one billion packages to be delivered in total.
At the end of the Aspara Conference, I went outside the venue to find the vending machine and tried for the first time to pay with my face. One or two seconds after you looked into the camera, your telephone number and Alipay account appeared on the screen. For a moment, I felt a bit scared as it gathered the data so quick, but the time and energy saved could be used elsewhere. The technology has been here for a while though.
Data, a word seemingly indifferent and detached, is embedded in our lives, comprehensively and even poetically. From the selfies you input to unlock the system, the spot you park your car, the bad traffic you unfortunately get stuck in to the package you pick up at the local delivery station. Using Daniel Zhang’s metaphor, “data is our petroleum and computing power is our engine.” With the massive computing power, data could directly be converted into an easier way of life.