Waic2022 丨 Behind AI, the landing process needs to be accelerated urgently

On September 3rd, the Fifth Artificial Intelligence Conference came to an end. This conference conducted intensive discussions on various breakthroughs in artificial intelligence technology. The exploration of the landing situation has also attracted widespread attention from industry people.

The 21st Century Business Herald reporters have learned that in recent years, applications have become the biggest bottlenecks for the development of artificial intelligence, but at the same time, it also contains a lot of opportunities. Therefore, the implementation of artificial intelligence technology through application traction has become a consensus in the current industry. Essence

"If a discipline is to obtain long -term vitality, it is the only way to closely combine with the industry." He Xiaodong, vice president of Jingdong Group, He Xiaodong, president of Jingdong Smart Products and Service Department, said in an interview with the 21st Century Business Herald. From the perspective, advanced technology will inevitably move towards the industry. In the past few years, the technology has not been mature enough, so the large -scale application prospects are not clear. But at this moment, artificial intelligence technology has become mature, and the timing of industrial landing is just around the corner.

Aiming at specific scenes

In the discussion around the Yuan universe, the integration of the Yuan universe and the major scenes has always been an unsophy topic. Different from the focus of the industry in the cultural and entertainment industry such as games, from the view of He Xiaodong, the value of the Yuan universe is not just the virtual NFT. Essence

"If we can achieve holographic human -machine interaction and accurate digital reconstruction and digital simulation, then the industry will usher in great efficiency improvement." He Xiaodong said that digital native technology is the most critical place in the Yuan cosmic technology. The actual landing scenario will help the target industry to achieve good cost reduction and efficiency. Taking the logistics industry as an example, the super intelligent supply chain depends on the ultimate simulation of each link of the entire supply chain, and the ultimate based on this is the ultimate. It is possible to reproduce the opportunity to become possible.

In addition to digital simulation, holographic human -machine interaction is also considered a new experience given by the Yuan universe to thousands of industries. In this regard, He Xiaodong believes that the ultimate human -machine interaction experience can be landed in the consumer -level application scenario, but in the game, in the game There are many places to apply. "For example, service, this service is a very extensive service. Where you need to consult, you need to provide interaction, emotional communication, you can have such a timely ubiquitous interaction."

Digital people are one of the most hopeful forms in the process of digital native and human -machine interaction and other technologies, but in the process of practical application, some digital people's movements are very rigid, and they can only play the effect of display. Actually handle business interaction.

He Xiaodong uses the digital people in the banking scene as an example that if the digital people want to handle the entire bank's business, they need to overcome several very important problems. The first is to avoid noise, because bank counters are relatively noisy places. Secondly, the bank digital person needs to be effectively connected with the background business support system. It must be able to solve the specific business problems of many banks, transfer, collect money, change the password, etc. This means The technical capabilities must reach very high standards, so the synchronization of these aspects is also a great challenge.

In the end, whether digital people can achieve high real -time, high flexibility and highly adapt to the landing scene, and to effectively interact with humans, it needs time to verify.

Data Smart Cognition

Outside the Yuan universe, data is the basis of artificial intelligence technology, and the focus of the industry around its own excavation and application is also the focus of the industry. From the perspective of Liu Yidong, the chief technical officer of Midu and the executive dean of the Micro -Heat Research Institute, this is mainly due to the complication and cross -platform characteristics of data in the current environment.

"At present, the data generated in the Internet does not exist independently, but at the same time mixed together. How to connect the information of multiple modes together, intelligently understand the association and content itself The technical ability is also the pain point in the industry. "

Intelligent understanding and cognitive technologies for multimodal content played a vital role in this. What is the meaning of so many modal content? Where is its specific application?

In an exclusive interview with the 21st Century Business Herald, Liu Yidong pointed out that intelligent understanding and cognition around multi -modal content is one of the main businesses of honey. He further explained that in general, the texts on the Internet include text and pictures on the Internet. There are various forms such as audio and video, and these are collectively referred to as a model in the field of content cognition. More accurate models and new methods for machines and people will be emphasized again, so it will provide better intelligent decisions to the government, enterprises and individuals in the future.

He also pointed out that the current multi -modal AI is constantly breaking the barriers of a single sensory, and the semantic information concept of multiple types of data is conceptual. The combination of industry knowledge and AI will also give birth to a new "knowledge calculation".

What is the problem of cost problems?

In fact, high costs are also one of the reasons why artificial intelligence technology is difficult to land. What is the solution of the enterprise? He Xiaodong told the 21st Century Business Herald that aiming at the common needs of customers to launch a standardized product, and then superimposed a personalized solution is an important means to solve the cost of cost in the industry. "JD artificial intelligence technology encountered a cost problem when it was landing inside. After exploring, we finally summarized a idea -forming an artificial intelligence industrialization platform, that is, as standardized as much as possible at the product level, adding a layer of application platforms in the middle. Customers. There are various customized needs, and the underlying technology customization This model is a high cost, so we want to encapsulate on the product layer. need."

He Xiaodong said that in the process of serving customers, JD.com tries to make standardized products as much as possible. Whether it is intelligent dialogue or smart production as much as possible to define a standardized product. "Used", or it can be applied after simple and fine -tuning. "

On this basis, JD.com further serves the personalized needs of all walks of life through the combination of the product, so that the cost can be effectively controlled while ensuring high delivery efficiency. "This method is equivalent A layer of packaging is added in the middle of the solution. This business model can make the technology and solutions decoupy as much as possible, so that the needs of technical iterative and user solutions will not be binded so strongly, so that customized customized A decrease, improve efficiency. "

For artificial intelligence companies with smaller scale and aiming at the market, standardized stacking specific solutions is also a solution to its cracking cost problem. Liu Yidong told reporters that honey not only uses AI capabilities to assist customers in work, but then there are many targeted services, which will perform some customized outputs in the service process of key customers. In this process, they will find and summarize the common needs of similar customers, and then form a certain standardized product.

- END -