In the past few years, with the development of the industry, the threshold of AI technology has been lowered, and we have entered the +AI era from the AI+ era. From the perspective of the development of the entire industry, the biggest opportunity we see in the next 10 years is the efficiency improvement of traditional industries, + AI empowerment, which will play an important role in restructuring and upgrading China’s economy.
When the rapid development of the Internet triggered industrial reforms, discussions on “Internet+” and “+Internet” emerged endlessly. Everyone must have heard a debate: What is the future of the Internet? Is it the Internet + traditional enterprises, or traditional enterprises + the Internet?
From 2012 to today, finally Internet + and + Internet are integrated. And now that artificial intelligence technology has come to be deeply integrated with the industry, what is “AI+”, that is, using AI plus traditional industries? Or “+AI”, that is, adding AI to traditional industries? Although the two will eventually merge, will the process be the same?
In the past few years, the development of AI has undergone a huge transformation, from technology-driven to commercial-driven, and the speed of industrialization and commercialization is getting faster and faster. For example, Transformer technology, from the publication of papers to the production of blooming applications, took only 2 years to complete the 30-year path of Convolutional Neural Networks (CNN) that year.
There are four reasons for the acceleration: First, the advancement of software tools, and the rapid maturity of deep learning frameworks; Second, hardware acceleration, making it easier to use; Third, cloud computing and other technologies have greatly reduced the implementation and deployment costs of AI algorithms; Fourth It is the emergence of AI talents in large numbers. These factors have pushed us from the era of AI+ to the era of +AI.
Difference between AI+ and +AI What is the difference between AI+ and +AI?
In the AI+ era, AI companies focus on technology and start their businesses with talented scientists as their core. There are very few such companies. After all, scientists who understand AI are limited. They are sought after by capital and become the first AI companies. Four or five years ago, with more and more talents who understand AI, tools became more and more popular, so more traditional companies began to think about how to integrate AI. Therefore, we have gradually entered the +AI era, that is, AI applications dominated by traditional companies.
Of course, in another five years, I believe that AI will enter the next stage-everywhere. AI applications will become simpler and simpler, and traditional companies can also use a simpler and more grounded model to introduce AI into the company, just as the state of IT today is the same. We can see a few more specific examples: AI+ companies focused on voice, vision, and chips in the early days. The +AI company focuses on retail, finance, manufacturing, transportation, energy and other fields, such as the cooperation between Wenyuan Zhixing and Guangzhou Baiyun Taxi Company.
According to PricewaterhouseCoopers (PWC) predictions, artificial intelligence will bring economic value of 100 trillion yuan to the world in 2030, and these values will be mainly created by the traditional enterprise + AI model. Why can traditional enterprises + AI create so much value? There are several important reasons:
First, the traditional industries are large in size, and the added value shows the effect of scale. For example, a bank or a car company, if AI can help it increase its efficiency by 3% or 5%, the value it generates is already huge.
Second, traditional companies have deep accumulation and high barriers to entry. AI practitioners may think that the technical threshold is the highest, but in fact, as mentioned earlier, the threshold of AI is gradually lowering. Now, it has become relatively easy for a bank to integrate AI, but it is very difficult for an AI company to become a bank.
Third, traditional industries can drive the fission of the ecological chain of technological upgrading. Traditional industries have formed a large-scale upstream and downstream ecology. Technological changes will affect the value of the entire ecological chain and bring about fission effects.
Fourth, traditional companies have different transformation needs and a high degree of customization. Although AI is powerful, its popularity is limited. Currently, no AI can be used directly as a platform. Every company needs to customize it to a high degree according to its own needs. For example, some unique data needs to be collected and cleaned up, and some companies may need to add more sensors.
Therefore, there is a big difference between the Internet and AI. The Internet can form a huge platform, and AI is more of a great technology that will empower existing platforms. Will AI itself produce a platform? I am not so optimistic. Therefore, AI+ will continue to be valuable, but +AI is a direction that is larger and contributes more to the society and economy.
Traditional enterprises consider +AI’s suggestions
What kind of traditional companies need to consider +AI at this stage? I have three suggestions:
The company is growing, and now it needs to expand or reduce costs, and it has business needs for its own development.
The company must have enough structured and massive data, and it must be related to business indicators. Combining with AI can create business value.
The company has a visionary and courageous CEO, has a culture that actively embraces change, and believes in AI empowerment and technological transformation.
Specifically, the introduction of AI in traditional industries can generate value from four aspects: ①+AI single link reduces costs and improves efficiency, and saves money with AI; ②+AI single link optimizes empowerment, and uses AI to simply replace a certain link; ③ +AI process intelligent empowerment, using AI to transform the company’s more important processes; ④ +AI reconstruction of the entire industry rules.
The cases we see today are more of the first three aspects, which are also the more classic traditional enterprise + AI model.
+AI single-link cost reduction and efficiency improvement, a subsidiary of Innovation Workshop, Chuangxin Qizhi, has made many products:
Quality inspection of clothing production line. Using machine vision to detect the size or damage of clothing, the detection accuracy is as high as 99.99%, which is 7 times more efficient than before.
Retail display quality monitoring. Use AI to check whether the display of goods on the shelf is compliant, and help brand owners do AI intelligence empowerment, which can achieve results in seconds, and the recognition accuracy rate is as high as 98%.
Assembly quality inspection of motor parts. Use AI to inspect the production process to see if there are any defects in the parts, such as whether the timing of the engine is aligned, the result can be obtained in 1 second, and the inspection accuracy is as high as 99.99%.
Intelligent monitoring of hot metal replenishment. Use smart methods to monitor whether there is an empty burning phenomenon in the molten iron production process in real time, remind workers to refill materials, and save electricity on a large scale.
+AI single link optimization empowerment
Give an example from the financial industry. Innovative Workshop Artificial Intelligence Engineering Institute conducted such an experiment: helping a very large loan company use AI to improve loan review, reducing the default rate by 14% and saving tens of millions of dollars.
+AI process intelligent empowerment
For example, smart operations in the retail industry. Use AI to help stores predict product sales, reduce stock-out rates, make inventory more intelligent, and connect logistics, warehousing, manufacturing and other links, so as to easily know how much each product should be produced, how much should be stored, and where to store it. When are they sold out, when they need to be restocked, and even where they are placed in the mall to optimize sales, etc.
+AI reconstructs the entire industry rules
The medical field is a typical example. We invested in an AI pharmaceutical company to use generative chemistry and anti-neural network technology to find the most suitable small molecules, optimize the drug discovery and production sequence, increase the probability of passing clinical trials, greatly accelerate the development of small molecules for new drugs, and enable the development The research and development speed is increased by 5 times, and the research and development expenses at this stage can be reduced by 3 to 5 times.
Innovative Works has invested a total of more than 40 startup companies in artificial intelligence. Recently, it has also discovered that more and more AI companies must land in order to be able to dock with traditional companies and cooperate with traditional companies to create maximum value.
Today, China is facing a very important moment. Traditional industries are facing many challenges, especially needing to reduce costs and increase efficiency. As a major manufacturing country in the world, we are now facing increasing labor costs, insufficient productivity and efficiency, and declining population and total factor productivity, making it difficult for small and medium-sized enterprises to survive.
Although we have a well-developed front-end, the efficiency of the consumer interface has been greatly improved, but the back-end is still lagging behind and lacks efficiency, which is extremely incompatible with the developed front-end. There are still many offline commercial formats that lag behind, and there are still 7 million traditional mom-and-pop stores in China. Many traditional industries urgently need to improve efficiency, such as education and medical care, and artificial intelligence can do it.
Therefore, from the perspective of the development of the entire industry, the huge value creation in the past 10 years has mainly come from front-end innovation. In the next 10 years, the biggest opportunity we see is the efficiency improvement of traditional industries, + AI empowerment. This is also the most promising direction for our investment.
In addition, there are two other factors that make us optimistic about +AI. First, although the new crown pneumonia epidemic is a disaster to the world, it has actually changed our usage habits, allowing more businesses to shift from offline to online, more data-based, and accelerating the implementation of AI.
Second, new infrastructure. Traditional enterprises embrace AI and need to have a very good foundation in computing, communication, and data. If you want to achieve +AI, data centers, 5G, IOT, and big data are all very important infrastructures.
Therefore, I believe that under the new infrastructure, +AI can achieve data, IT, and cloud in one step, and it will play an important role in restructuring and upgrading China’s economy.