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

Industry AI, industry AI, industry Internet, these words have become popular words today.

After almost eating up the flow dividend, technology companies need to move towards the industry, which is the consensus of most technology companies today. However, a core difference between the industrial market and the consumer market is that there are too many differences between each industry. The requirements of the energy industry are obviously different from those of the education industry, and a set of smart technology solutions that dominate the world is obviously unreliable.

If we pay more attention to the trend of the industry AI, we will find that since the second half of last year, more and more words such as industry expert and Know-How are being mentioned in the PPT of related projects.

When AI began to emphasize the importance of Know-How and called on companies with Know-How to participate in the progress of AI industrialization, maybe we should comb through this problem from scratch.

How do industry experts cooperate with providers of AI algorithms and computing power, and what obstacles does the lack of industry experts bring to AI's entry into the vertical industry?

Let's take a look at the mysterious Know-How, the dream lover of the big factory and the lifesaver of the startup company in the AI circle.

The Intangible Wall of Industrial AI

The so-called Know-How refers to the "secrets of the industry" taught by those masters to their apprentices in the craftsman era.

In the era of large-scale mechanical production, Know-How seems to be more and more transparent by information explosion. However, under the situation of increasingly refined industry segmentation and many industries climbing the science and technology tree, Know-How continues to precipitate and accumulate in the economic entity.

For example, cars and ships are industries with a large number of Know-How nodes. Even when the basic technology is not difficult to copy and the industry supply chain is relatively transparent. The level of production of cars and ships is still difficult to replicate, in part because there are too many "secrets".

Know-How can be understood as a kind of ability, a kind of resource, or someone who is called an industry expert. In the investment industry, Know-How is also seen as one of the scoring criteria for a start-up project-if the founder of an auto e-commerce is an old Youtiao who has been immersed in the auto industry for decades, then BP tends to say that we have Know-How.

When the intelligent technology of data and network dominated by AI hopes to enter the industry, Know-How will become extremely important.

One of the essential reasons why the so-called industry AI or industry AI can improve labor productivity is that machine learning technology can be used to realize data analysis and re-mining, so that AI can reintegrate the data of the original extensive growth, and re-obtain some optimal solutions of industrial efficiency. For example, calculate the proportion of raw material input, the law of storage and placement, industrial process reshaping and so on.

In addition, another function of AI is to provide voice and vision capabilities for the industrial end, such as campus voice navigation, machine vision-based quality inspection, and so on.

To make a point, these AI capabilities need to move from all aspects of complex details into existing industrial entities. But exactly how to get in, what unpredictable issues you need to pay attention to, and when to recoup the cost of technical iteration-- these answers are in Know-How 's hands.

With the AI of algorithm and computing power as well as PPT, it is inevitable to encounter this kind of awkwardness when entering the subdivided industries, especially the industries with relatively strong industrial attributes. Although AI sounds reliable, it is impossible to move forward without the help of the Industrial leading Party.

A more obvious problem comes from the talent reserve.

Generally speaking, AI algorithm engineers pay attention to the training deployment of deep learning and other related content. Real logic, detailed AI solutions, cost-effective estimation of enterprises, and flexible growth of industrial intelligence are not part of the daily consideration of algorithm architects or AI developers.

On the other hand, industry experts know the industry cycle like the back of their hand, but it is difficult to have the experience and opportunity to learn and understand AI-related content. Eventually, the industrial AI has become two neighbors who speak their own words and are difficult to understand each other.

In comparison, today's AI this end is relatively transparent, the real industrial chain cooperation pressure, came to the AI company looking for industrial Know-How side.

In many actual AI industry integration cases we know, we will find that the problems found by industry experts are not in the conventional vision of AI and data intelligence technology. The discovery of a good problem often indicates the opening of a new industrial space.

In the final analysis, the lack of professional knowledge and professionals is becoming an invisible wall that limits the market of AI landing industry. This scarcity is neither a technical problem nor a market problem, but it actually restricts the pace of AI.

How Know-How works

Ideally, a qualified Know-How or Know-How company is needed to help when machine learning and other technologies enter a factory or an enterprise. In order to ensure that the general AI technology and differentiated enterprise needs to achieve docking.

1. Find and control the industry differentiation in AI work. The working mode of machine learning is to extract abstract features and transmit them to the machine in reverse, so as to realize intelligence. However, it is difficult for AI developers to predict what features to extract, what are the problems in the extraction process, and what are unreasonable in the work. For example, the famous AI to improve the rate of good products, in the end what is good products, the definition of each industry is different. This definition is the differentiation node that Know-How needs to provide.

2. Key training data. AI is inseparable from data, but although there are many general data, the direction is relatively thin, often lack of the actual potential of industrialization. And where is the undisclosed industry value data? This is also the value of Know- how talent and the company.

3. The understanding of cost and value. Using AI always sounds good, but how much manpower and material resources should be invested in this valuable thing, when to recover the cost, and how much value can be created in the future all depend greatly on the profit ratio of the industry. It is the responsibility of Know-How to estimate the entire input-output cycle for industry users.

4. The understanding of industry chain. Today, there is also a situation, that is, their own enterprise system AI, production capacity has increased, but the ability to connect with suppliers has been weakened. In the complex industrial chain, the renewal of an enterprise from management system, operation and maintenance system to production system will be affected and subject to the relationship between the upstream and downstream of the industry. The understanding and prediction of these relationships is very important for enterprise technology decision-making, and its grasp ability is also in the hands of Know-How.

In this way, it seems that Know-How is a bit like an intermediary between AI and industry. Most of the time we don't want to find an intermediary, but we find that if we don't find them, there are more problems and less efficiency.

So for AI, who is today's Know-How?

Who is the quasi-Know-how that AI needs?

Those who have access to the core data of the industry and understand the technical system, supply and demand of the industry, departments and third-party enterprises are obviously the important task of leading the party in AI.

In general, there are several types of Know-How that can be used by AI to form close ecological alliances to build channels for AI to enter the industry:

1. The IT department of the enterprise. The IT department of a non-Internet company often looks like it is in charge of repairing the network. However, after a long period of training, similar departments often accumulate a unique understanding of the needs of the industry, and accumulate a large number of key data that can be used by machine learning systems. When an enterprise begins to use AI technology to expand its production system, it is often dominated by the IT department.

2. The backbone of technical business. In many real economy and traditional enterprises, there are very senior and high-level technical professionals. They not only have the knowledge of their own industry, in fact, they often pay attention and enthusiasm to new technologies, but also have the ability to relearn. The full use of these talents can be used as the key for AI to enter the industry to solve the problem of integration. At the same time, taking the technical backbone of the industry as the audience and training advanced industrial AI integration talents has also become the goal of the science and technology giants' attention.

3. Mature technical service providers in the industry. In the fields of automobile, energy, metallurgy and so on, there seem to be a large number of technical service providers with global coverage. Behind the giants, there are thousands of specialist solution providers working. Taking these supply chain enterprises as a breakthrough point to release AI dividends can be used as the development model of AI in many industries.

4. Data and consulting service providers. On the other hand, there are a large number of strategic consulting and industrial data service organizations in many industries. These enterprises around the production data release value, accumulated a large number of enterprise needs and real ideas. If it can be used effectively, it can also be used as an exit for AI to dock with enterprises.

Of course, there are a lot of identities and possibilities for Know-How. But overall, finding these people and taking advantage of them is just the beginning in today's AI world.

Service enterprise is a very difficult and changeable market. It is an inevitable trend for AI to awaken some helpers first.

What does the scarcity of Know-How lead to?

Technology giants have noticed the importance of Know-How, began to build their own Know-How ecology, coupled with the industrial AI proposition is actually in the initial stage, the acceptance of the industry is still very limited. Let the Know-How providing services for AI as a whole in the stage of supply less than demand.

The scarcity of Know-How makes the development of industrial AI inevitably experience the following trends:

1. Key industries have begun to take the lead in developing AI. We are all talking about AI entering all walks of life. Obviously, however, it is impossible for AI to enter a hundred industries at the same time. Industries with high degree of digitalization, sufficient Know-How resources and good IT foundation are more likely to take the lead in developing industrial AI ecology. At present, several industries such as automobile, retail, manufacturing and Internet have a better AI foundation.

2. AI will face a very complicated situation when it enters the vertical industry, especially in the industrial field. Due to the complexity and wide differentiation of the industrial field, the cycle and cost of AI technology entry will be very high. The complexity of Know-How also makes it difficult for the industry to have an intelligent process of cutting the mess with a sharp knife, which can only be plotted slowly.

3. Owning Know-How will become the trump card of some kind of AI startup. Today's AI start-up companies, more is to fight Daniel, spell algorithm uniqueness. These stories will become the sexiest part in the eyes of investors, and after the entry threshold of vertical industry is constantly exposed, the level of Know-How owned by enterprises will begin to affect the financing ability and development level of AI startups. At the same time, having the ability of Know-How will also become an ecological bargaining chip for startups, technology giants and computing providers.

4. Telling cases and storytelling will become extremely important. Finding Know-How is a very different and complicated task, which makes enterprises realize the existence and rationality of industry differentiated AI solutions, and it is also a complex task. In this case, the choice of AI enterprises can only be to cherish the existing cases, disassemble the logic, strengthen publicity, make more industrial relations aware of the possibility of cooperation, and strengthen their own active attraction of Know-How. Therefore, AI enters the industry this cycle, basically must be the case is the king.

From algorithm problems, computing power and data problems to Know-How problems, in essence, AI is moving towards the mysterious industrial world step by step. In essence, AI is a new technology that will directly affect the steps of industrial production, just like coal and electricity.

To let people who understand AI and people who understand the industry know each other as soon as possible, and even form mutual attraction, is an indispensable link in the development of AI.

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.