Intelligent + era artificial intelligence development three elements need to be upgraded
Release date:



 As we all know, the development of artificial intelligence is inseparable from three major elements: algorithm, computing power, big data.

    The logical relationship between the three complements each other. I believe everyone knows: In 2006, Professor Hinton proposed a deep learning algorithm, which opened up an innovative breakthrough in the AI theory community. The follow-up of the mobile Internet has promoted the conditions of AI development. The production. The combination of big data and deep learning algorithms is combined with the power of Moore's Law to quickly increase the output, and output AI solutions in different scenarios and industries, such as face recognition, object recognition, and speech recognition that are often seen today. , speech synthesis and many other applications and results.


    Artificial intelligence has evolved into a small monster that is being fed by big data, and the ability of artificial intelligence to achieve self-learning seems to be far away in the absence of a significant breakthrough in deep learning algorithms, so the alternative to AI for humans. As well as threats, it is far from reaching the point of worrying humans. The most widely discussed is the application of artificial intelligence in various fields.

    Especially during the two sessions this year, artificial intelligence has appeared in the government work report many times. This year, the concept of 'smart +' was put forward in more detail. The government work report puts forward 'smart +', that is, 'energy' production scenarios through artificial intelligence technology, creating an industrial Internet platform, improving production efficiency in all walks of life, creating new market demands, and becoming a new driving force for China's economic growth.

    When the sexy buzzword of artificial intelligence collides with the heavy and dull words of industry. The problem is coming:

    Under the trend of development, will the three elements of artificial intelligence still apply?

    If we pay more attention to the trend of industrial AI, we will find that since the second half of last year, the PPT of related projects has increasingly mentioned the words of industry experts and Know-How. From algorithmic issues, computational and data issues, to Know-How, AI is moving forward to the mysterious industrial world.

    So what exactly is Know-How?

    The content of Know-How excerpted from Baidu Encyclopedia is as follows:

    Chinese translation: Technical 诀窍, refers to the general name of the skills taught by the masters of the medieval hand workshop to the apprentice. Now more refers to the technical know-how and expertise required to work in an industry or to do a certain job. Take the automobile manufacturing industry, if the manufacturing of a car is like cooking, and the technical documents are like the “secret recipe” of the food, then Know-How is the process of “single secret recipe”. The 'world famous chefs' have been repeatedly tried and tested over a long period of time. Know-How is the 'fishing' of the ancient Chinese teaching that 'giving people a fish is not as good as giving people a fish.'

    Ideally: If artificial intelligence is to enter an industrial field, it needs a professional industry expert to guide. This expert can be a person or a company. Know-How is more like the existence of this intermediary mechanism to ensure AI's applications and needs in various fields are fully docked and satisfied. Its benefits are as follows:

    1. Control industry differentiation to ensure industrial application.

    The principle of deep learning is to extract the abstract features of the big data marked by humans and deliver them to the machine in reverse to achieve intelligence. But what features are extracted, what issues to pay attention to during the extraction process, etc. These are things that AI cannot predict and estimate. For example, AI raises the yield rate problem, what is the definition of good product? Weight or surface texture, or running amplitude. A definition similar to this feature is the Know-How point of AI in industrial applications.

    2. Acquisition and application of key training data.

    AI is inseparable from data. Nowadays, the application of artificial intelligence development is concentrated on general data. With the trend of information and information fusion between the Internet of Things and the information network, the value density of data is greatly increased, and even the trend of edge computing is triggered. The acquisition latitude of big data. And the increase in density, challenges and opportunities coexist. First of all, in the industrial field, how to apply the undisclosed industrial value data, how to prevent leakage, and secondly, the industrial application in the trend of information fusion, may also find some new data latitude and value, how to mine, how to apply. The above two points can be understood as another Know-How value point of AI in industrial applications.


    1. Try to understand and sort out the relationship between cost and value in the supply chain.

    AI empowerment, it sounds very good, but in the end, how much manpower and material resources need to be invested, what time can recover the cost, and how much value is expected to be created depends on the stable cost-value relationship formed by the industry for many years of application. Estimating the input and output cycle of the entire AI empowerment for industry users is also the value point of Know-How.

    2. Add complex business game logic and relationships to the industry chain empowerment.

    AI empowers the industry and its production capacity has gone up, but it is possible that the ability to connect with suppliers has weakened. In a complex industrial chain, a company is closely intertwined from the management system, operation and maintenance system to the production system, and the upstream and downstream depths. Any change in one dimension may affect the upstream and downstream relationship, and the change conditions are also subject to Downstream relationship. The understanding and prejudgment of these relationships is also the value point of Know-How.

    In fact, the Know-How ability is really a bit like an intermediary between AI and industry. Just as we went to buy second-hand houses, although we didn't want to find an intermediary, we found that it was really impossible to leave them. To understand this, the application of smart + and the promotion of landing will require more rationality in it:

    Smart +, can't be quick and confusing, only step by step.

    If AI is to enter the vertical industry, especially in the industrial sector, it will face a very complicated situation. Don't be hurry to find fast, because of the complexity and differentiation of the industrial field, the cycle and cost of AI empowerment will be very high. Each industry is its own Know-How point, which needs to be merged one by one and broken one by one.

    Smart +, the entry point should give priority to certain key industrial areas, such as the new 100 billion market rail transit field

    The industrial field is a Kind-How point-intensive area, but it is also the area with the most obvious driving effect of industrial clusters. It has strong purchasing power and market capacity, and if some industries are highly digitized, then its The innate data advantage reduces the cycle, cost, and risk of AI empowerment. For example, in the rail transit field represented by the Guangzhou Metro, the new infrastructure proposed by the Central Economic Work Conference at the end of 2018 specifically proposed a new type of infrastructure construction such as rail transit, which is a new model for maintaining growth and stabilizing innovation. . The planning outline of the Dawan District of Guangdong, Hong Kong and Macao also clearly stated that it is necessary to establish a “one-hour living circle”. This has great value for the interconnection of the entire Bay Area, whether it is the exchange of logistics, business flow and capital flow. One hour of life circle construction, the best mode is rail transit. At the same time, the rail transit traffic field has a huge potential for market space: Guangzhou announced that the rail transit industry development three-year action plan specifically mentioned that by 2021, only the rail transit industry in Guangzhou is expected to reach 120 billion yuan, realized in 2023. Strive for an output value of 180 billion yuan. In terms of digitization, rail transit is also at the forefront. In January 2019, with the technical support of Jiadu Technology, the Guangzhou Metro Line was the first in China to support all the gates to support the pass code (Guangzhou Metro APP ride code, Guangzhou Metro ride code Alipay small program, Guangzhou Metro ride code) WeChat small program), while Jiadu Technology also realized the face recognition, fuzzy image processing, thermogram number statistics, behavior analysis and other technologies in the subway intelligent system by means of its own Know-How ability in the industry. Features. According to its announcement, in March, Jiadu Technology also won the bid for the 11.8 billion Guangzhou Metro New Line Project, which is the application of a batch of AI new products such as fully automatic operation system, face recognition ticket inspection system and digital station-based intelligent platform system. It will lay the foundation and foundation for digitalization, networking and intelligence in the field of rail transit.


    With the ability of Know-How, it is the key to intelligent + landing!

    In the process of technology and scene integration, the core technology and industry Know-How will become the key to building competitive barriers;

    Whether it can solve the pain points of the industry and promote the real cost reduction and efficiency increase of the industry is related to the survival and development of AI enterprises. With Know-How, it will become a kind of AI company's card. Today's AI company, more is a big cow, spelling algorithm is unique. These stories can be told to investors, but when they are applied in the real industry, whether or not they have the Know-How level will affect the financing ability and development level of AI.


    Therefore, AI can enter everyone's life like water and electricity, but it does not necessarily enter every factory like water and electricity.

    Let the people who understand AI and the people who understand the industry meet each other as soon as possible, and form a close cooperative relationship. This is the development of artificial intelligence to the present day. In the era of intelligence + new era, everyone needs to consider the issues.

    Why is this the fourth industrial revolution?

    Because industry is the foundation of a country, it is the source that affects hundreds of industries.