Shulou(Shulou.com)06/02 Report--

Does AI work?

In this year, which takes the concepts of AI industrialization and AI business landing as key words, countless enterprises, large and small, are answering this question by themselves. When people are talking to smart customer service on the phone, and when the smart cameras on the production line are sorting products, we seem to have heard the answer.

But for those who work hard in the industry, getting the answer to a technology that has been dug up by academia since the 1950s is not as simple as it seems.

This is especially true for AI practitioners like Zhou Bowen who have dabbled in both academic and industrial circles.

JD.com Group announced the establishment of Jingdongyun and AI Business Department to integrate the structure and responsibilities of the former three business departments of JD.com Yun, artificial Intelligence and IoT, headed by Dr. Zhou Bowen, Vice President of JD.com Group, and report to Mr. Liu Qiangdong, Chairman of JD.com Group and CEO. JD.com 's cloud, AI, and IoT three major technology modules are in line with each other, speeding up progress on the road of technology industrialization.

For Zhou Bowen, it is now halfway up the mountain.

He is not a typical scientist.

When it comes to Zhou Bowen, who joined JD.com in 2017, I believe anyone who follows AI or the technology industry can blurt out a series of labels on him: chief scientist of IBM Watson Group, former head of artificial Intelligence basic Research Institute of IBM Research headquarters in New York, IEEE Fellow, 20-year practitioner in AI field.

Like many technology companies, JD.com recruited a scientist.

But unlike many science and technology enterprises, as an "atypical" scientist, Zhou Bowen has a long-term pursuit of technology industrialization and commercialization behind a series of labels.

For example, in 2003, when mobile devices were not yet popular, Zhou Bowen developed the world's first fully embedded two-way real-time voice-to-speech translation system with large vocabulary, and deployed this system on the handheld end. In that era when the mobile Internet was not yet popular, for people living in a foreign environment, products that could be carried with them and solved language problems to a certain extent really helped them a lot.

A technological innovation can meet people's real needs. Since then, Zhou Bowen began to focus on the commercialization of technology. In the interview, he said something like this:

"so I always think that the purpose of the technology core is not the paper, the paper is the means of presentation, and the ultimate goal is to create value."

So in the next ten years of Zhou Bowen's career, we can continue to find traces of his balance between basic research and commercialization of technology. For example, in 2014, he took the initiative to join the export IBM Watson Group from the IBM Research, which pays more attention to basic technology research, to explore the commercial and platform model of artificial intelligence.

It can be said that although AI really began to mature and applied to the industry on a large scale in the past five years, scientists such as Zhou Bowen have already taken advantage of the morning light.

From the technology trial field to every small change

In this way, it is not difficult for us to see the correspondence between Zhou Bowen's technical proposition and JD.com 's actual situation.

Zhou Bowen's technical proposition is to make technology platform and commercialized to give full play to value, while JD.com is characterized by a wide range of business scenarios, e-commerce, logistics, finance, customer service, supply chain, insurance, logistics real estate. These business scenarios accumulate a large amount of real data and become the best divergence field for technology. In addition, JD.com itself has a strong open platform attribute, exporting e-commerce, finance, logistics, advertising and other services to users and partners, it can be said that the channel has been opened, only waiting for the technology flow to wash away.

Such an enterprise can be said to be the trainee of natural technology, but also the natural exporter. Perhaps because of such natural conditions, JD.com did not blindly choose those scholars who could brew technological breakthroughs in the next century and hit it off with Zhou Bowen, who focuses on turning technology into products and value.

We can witness JD.com 's transformation through a series of figures: JD.com invested more than 13 billion yuan in technology in the first three quarters of 2019, more than in 2018; and JD.com Cloud has improved a lot in terms of the richness and stability of its technology products. Statistics from IDC show that JD.com Cloud's public cloud business grew by more than 500% in the first half of the year. JD.com AI's new generation intelligent supply chain platform with NeuHub as the core carrier has also been selected into the intelligent supply chain national new generation artificial intelligence open innovation platform announced by the Ministry of Science and Technology, making JD.com a member of the national team.

These numbers seem cold, but they really change people's lives.

For example, Qian Aunt fresh Meat Store in Guangzhou vegetable Market, which has become an "online celebrity shop", an innovative retail store. With JD.com 's analysis and learning of online and offline data, we can also realize the intelligent management of the supply chain. The time, place and quantity of purchase, and the time of discount and promotion can all be made with the assistance of AI. The low-cost realization of "not selling overnight meat" has led to the chain development of this small store, and even achieved a turnover of 60 million this year.

For example, after the application of intelligent customer service, during the peak period of JD.com 11.11,618, not only the needs of users can be met more quickly, but also the burden of customer service staff is relatively reduced. During the promotion period, when Zhou Bowen was on a "rotation" in the customer service department, his customer service colleagues told him that compared with previous years, there was not enough time to eat, and this year, with the help of AI customer service, he finally had plenty of time to eat and go to the bathroom.

The successful commercialization of technology not only brings economic benefits or national recognition, but also changes every ordinary life, so that small and micro business owners can also have big dreams, so that customer service workers can have a safe meal at JD.com 11.11. Such a real change is virtually consistent with JD.com 's concept of "trustworthy AI".

Surgical Reform of typical Science and Technology Enterprises

Behind these changes, we can see the technological platform of a "typical" technology company.

What is a typical technology company?

Such companies often have no shortage of technical personnel, and there is a lot of technical support behind their business. However, the mobility of technology within the enterprise is very poor, specifically reflected in that technology An in business A may also play a role in business B, but often the actual situation is that B department will re-develop technology B which is very similar to technology A.

When Zhou Bowen first joined JD.com, he spent a long time collecting the technical requirements of various departments and promoting the restructuring between business and technology. Then it is screened by two criteria-one is the reuse rate of the technology, and the other is the sedimentation of the technology. Choosing the technology with high reuse rate means that the cost of repeated research and development can be reduced internally, and it also means that the landing application space can be found externally. As for precipitability, it means that whether the technology has the prospect of self-driving development can be improved with the help of real-world feedback as it continues to land.

Under these two standards, JD.com 's previous technical capabilities continue to emerge from the business itself, with an open space for development and progress. JD.com AI open platform NeuHub is a typical development achievement.

At the same time, JD.com is also strengthening technological infrastructure, such as the Technical Committee led by Zhou Bowen, which is responsible for attracting and promoting technical personnel, as well as sorting out JD.com 's overall technological ecology and strategy. Since Zhou Bowen joined JD.com, he Xiaodong, Mei Tao and other technical personnel have been recruited.

JD.com experienced a "surgical" change in the past two years, dividing the originally intertwined business and technology into suitable nutrition delivery. It is believed that this process can provide reference for many similar technology enterprises.

A person, halfway up the hill of an enterprise

So, what about after this operation?

While thinking about the commercialization of AI technology, we can not ignore the core attribute of AI technology. There is no doubt that the core of AI lies in the algorithm. It is precisely because there are powerful models that data can be entered into it and the results can be obtained by reasoning. However, the existence of the model often needs the computing force as the carrier, and the computing power exists either in the cloud or at the edge chip end. Furthermore, it is the devices that help AI collect data and feedback results, from the most common mobile terminals to sensors and cameras used in different scenes.

As Zhou Bowen summed up the technology strategy: "ABCDE" theory, namely AI (artificial Intelligence), Big Data (big data), Cloud (Cloud Computing), Devices (IoT) and Exploration (Frontier Exploration).

He uses an analogy to describe the relationship between ABCDE: business is very symbiotic. If you think of the whole technology as one person, AI is the brain, and IoT is the perception and collection of nerve terminals, and the execution of signals. Clouds provide the torso, muscles and blood vessels of the body, in which big data is oxygen.

For technology companies, cloud and IoT are indispensable deployment ways for AI to play an industrial role. Especially in the near future, with the popularity of 5G technology, data transmission will also drive the change of technology application form. Therefore, JD.com 's integration of AI, cloud and IoT is an inevitable trend.

Interestingly, when Zhou Bowen promoted the commercialization of IBM Watson in 2014, he also combined Watson with the IBM public cloud platform. The cloud has always been an important carrier of AI, and it is the same a few years later.

We might as well open our minds and speculate on what measures JD.com will take in the commercialization of technology in the future after the integration of cloud, AI and IoT.

1. Integration and opening up

One of the major characteristics of JD.com 's technology export is that technologies such as intelligent supply chain are often faced with physical industries or even small and micro physical industries. These users often have a weak technical base and require suppliers to provide as complete a solution as possible rather than just technical API. After the integration of cloud and IoT, JD.com 's future technology output may be more integrated, that is, "from integration to integration and opening up" proposed by JD.com Group, to package technical goods more perfectly and reduce deployment costs by opening the "box".

2. Collaborative research and development

After integration, the scope of technological precipitation can also be wider. For example, the research and development of algorithms can take into account the needs of different deployment models. In warehousing scenarios with relatively poor network conditions, edge deployment is more often used to enable cloud services, AI algorithms and IoT terminals to cooperate with each other, giving different users more detailed technical solutions.

3. Broad space for innovation

When strengthening basic technological research and integrating multiple business units go hand in hand, the two initiatives will eventually meet. It is not difficult for us to predict that in the future, JD.com will have more basic breakthroughs rather than new technologies born in the current business scenario, which will be deployed to real scenarios through cloud services or IoT products.

For a huge technology incubator like JD.com, it took only half a way to complete the transformation from retail to technology platform in two years. How to maintain the self-driven development of technology after the identity change, and how to face the competition brought about by the collective shift of science and technology enterprises to the B-end market, is the real best part of JD.com 's story.

For Zhou Bowen, a technology leader who has worked in the commercialization of technology for more than a decade, the era of AI industrialization is not a long-awaited Mid-levels scenery for him.

Up the hillside, there is boundless scenery.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.