The eve of every industrial revolution is often long and tortuous.
From the manually operated Jenny spinning machine to the coal based Watt steam engine, it took decades. If we want to reach the stage of large-scale and industrial application, it will be almost half a century.
Back in the past, in the first three scientific and technological revolutions, steam power, internal combustion engine power, and electricity have all gone through a long period of trial and error and growth from the emergence of such core production factors to the transformation of leading production relations and even industrial structure. In the first two revolutions, fossil fuel-driven machinery was the core factor, and the third revolution was the arrival of the era of electric catalytic information.
Nearly half a century later, with the improvement of the information and communication infrastructure, the transformation of the traditional Internet industry began to accelerate, and the penetration of the digital economy, represented by AI, into the traditional industry has deepened. When AI’s driving force for technological development becomes more obvious, it will be closer to the final arrival of the next industrial revolution.
So far, the proportion of AI in the output value of the Internet industry has also increased rapidly. According to iResearch Consulting, the global AI industry scale will reach 361.9 billion US dollars in 2021, of which China accounts for 404.1 billion yuan, accounting for nearly 16%.
In the face of this “gold mine” that has not yet been fully reclaimed, various technology companies are also accelerating their “rush to the beach”. For them, to bet on AI is to bet on the next era – who can make more progress in the field of AI, who can gain more voice in the next era, but this is not a company’s “one-upmanship”, but the “ecological reaction” formed by numerous AI technology companies.
At the sixth “iFLYTEK Global 1024 Developers Festival”, it may be an opportunity to take the latest achievements of AI technology such as virtual human, robot dog, and industrial platform as the guide, and the ideas of various AI-related fields as the medicine to jointly find the best path to the universe, digital economy, industrial Internet, and digital low-carbon.
But this is obviously not the end. How to exploit this “digital gold mine” through AI technology has undoubtedly become the common proposition of all technology companies at present.
Time, choose AI?
From the Internet, WWW, to today’s mobile Internet, it is the process that the Internet gradually deepens the transformation of social form. It is also the process that AI based on algorithm data and autonomous training model gradually moves from behind the scenes to the front.
For example, in the era of PC Internet in the first decade of this century, tool-based software and web applications and client-side online games dominated the mainstream. At that time, the market demand could be met through the back-end “heap technology” and the front-end “heap manpower”. AI and other AI technologies, limited by hardware computing power and programming level, have not yet occupied the mainstream in the Internet industry.
Since 2013, with the popularity of mobile terminals such as smart phones and the emergence of mobile applications such as today’s headlines, content recommendation through algorithms has gradually begun to take a strategic position in the field of consumer Internet, and its application scope has gradually extended from content to e-commerce, consumption and other fields. With the popularity of 5G and other communication technologies, the penetration of the industry chain has also deepened.
According to the 2022 China Digital Economy Development Report released by the China Academy of Information and Communications, the scale of the domestic digital economy increased from 22.6 trillion to 45.5 trillion in 2016-2021, more than double the expansion of the initial stage of the “13th Five-Year Plan”, accounting for 39.8% of GDP. According to the data of iResearch Consulting, the industrial scale of AI is only 404.1 billion yuan, only a fraction of the overall output value of the domestic digital economy.
This is not bad news. Because only the economic form after preliminary digital transformation is the “soil” suitable for AI technology to take root and sprout. In the long run, the industrial scale of this part of the digital economy is a blue ocean market that needs to be reclaimed by AI technology.
But from an objective point of view, the domestic AI industry still has shortcomings in terms of technology.
First, China’s basic innovation investment in the field of artificial intelligence is seriously insufficient. From the perspective of enterprise R&D and innovation, the innovation R&D expenditure of Chinese AI enterprises is still far behind that of the United States, Europe and Japan. Only in 2018-2019, the R&D expenditure of American AI enterprises accounted for 61% of the global S&T expenditure. Although the R&D expenditure of Chinese AI enterprises increased rapidly, with a growth rate of 34%, the actual share of global S&T expenditure was significantly smaller than that of the United States.
Second, from the perspective of AI intellectual property ownership, the total number of AI patents owned by various entities in China is more than 30000, ranking first in the world. However, most of the AI-related patents owned by Chinese enterprises are utility model patents with low threshold. Invention patents only account for 23% of the total number of patent applications, and most of the design patents will expire in five years. Whether to pay maintenance fees is still unknown.
However, due to the significant regional differences in domestic economic development, there are still many traditional enterprises that have not been digitally transformed, and there are generation differences between competitors with relatively high degree of digitalization. Even for companies with better digitalization in the market, their digital applications need to be updated timely to meet the new market environment, which also provides a large number of potential customers for domestic AI manufacturers.
AI technology, go to the front
Like all emerging technologies, AI technology is very “simple” in its early development stage, with algorithm intelligent recommendation or facial recognition technology as the mainstream. Today, relying on the development of communication technology and underlying hardware, AI has gone from consumer internet to industrial internet with the pace of 5G, and started to transform traditional industries from the outside to the inside.
For example, in coal mine machinery and equipment, the mine hoist integrates “machinery, electricity and fluid”. Its lifting energy and operating performance determine the production capacity of the coal mine. It is an important transportation equipment to ensure the safe production of the coal mine, known as the “throat of the coal mine”. Therefore, the safety of such key devices is essentially the safety of production personnel and plant.
However, depending on human and experience detection, there must be a dead corner between time and space. In Dingji Coal Mine of Huaihe Energy, the Antelope Industrial Equipment Guard Platform has realized the accurate identification of more than 14 kinds of faults of key components such as fan motors and bearings. The spot check load has been reduced by 80%, and the maintenance load has been reduced by 30%. It has solved the problems of safety accident risk and high operation and maintenance costs of customers’ key equipment.
The data center, represented by the antelope industrial equipment guard platform, is also one of the main ways for AI to carry out digital transformation of traditional enterprises. According to statistics, at present, the antelope platform has settled 226000 users, more than 455000 platform service enterprises, the total transaction volume of the platform exceeds 2.94 billion, more than 80000 student makers, 700+achievement cases, and more than 70000 scientific and technological achievements.
At the sixth “iFLYTEK Global 1024 Developers Festival”, Liu Qingfeng, chairman of iFLYTEK, also revealed that it is expected that the number of users of the Antelope platform will reach millions by this Spring Festival. For small and medium-sized enterprises, the Antelope platform can be enabled in business scenarios such as office and marketing through AI in the form of SaaS services.
For large enterprises, they can help enterprises reduce the burden and increase efficiency through industrial brain, artificial intelligence+RPA, and industrial process automation.
Compared with AI technology in the laboratory, AI technology applied in enterprises is obviously more sexy.
For example, in the daily inspection of traditional enterprises, robots equipped with AI technologies such as autonomous navigation, adaptive gait and multimode perception can not only be used as a supplement to human inspection, but also play a more important role in some scenes than humans.
Take the intelligent robot dog “Xiaohei” launched by iFLYTEK as an example. With the support of the above three capabilities and the acoustic imager, this kind of robot dog can be competent for the acoustic local discharge fault patrol inspection of substations and gas stations, and realize the partial replacement of human patrol inspection. At the same time, its small size can also go deep into some dead corners that are hard to reach by human, and achieve troubleshooting in some extreme scenarios.
According to relevant research reports, the scale of China’s intelligent robot market will be close to 100 billion in 2025. At the same time, the advantages of industrial robots in terms of efficiency, accuracy, quality and safety can effectively alleviate the difficulties caused by the rising labor costs and the inability of production capacity to meet the demand in time.
Intelligent manufacturing is the core of “Industry 4.0”. As an important carrier to promote intelligent manufacturing, robots will play an important role. In this process, AI, as the underlying technology, also grows with the development of the industry.
In the field of pan-entertainment, virtual digital people represented by virtual idols have sprung up in recent years. Last year, iFLYTEK put forward the strategy of virtual human interaction platform 1.0, which has begun to take shape today.
Most of the previous virtual people completed the shape of the virtual human through CG modeling, and then realized the driven virtual human through the joint dynamic and surface capture technology of human in the middle. The virtual idol group “A-soul” belongs to this category.
However, this kind of virtual human generally has problems such as high labor cost, high technical threshold and long production cycle, which can be applied to some application scenarios of virtual human, but it is difficult to meet multiple application scenarios through single technology and traditional dynamic capture effect in the future.
The virtual human generated based on the virtual human interaction platform 1.0, in addition to being close to the real person in terms of sound, shape, action and other factors, also combines the cognitive psychology theory to design the virtual human interaction emotion, and uses the massive text unsupervised semantic model to classify and predict the emotion of various texts.
The personalized 2D/3D virtual human construction system launched by iFLYTEK can generate and drive 2D and 3D virtual human in seconds with very few input conditions, such as one photo. There are mature cases applied to culture and tourism, finance, retail and other industries.
In this “iFLYTEK Global 1024 Developers Festival”, 468 teams participated in the virtual human image contest alone, and 44 excellent works were presented in full size at the exhibition’s cultural and creative market. Yu Jidong, president of iFLYTEK’s consumer business group, said at the developer’s festival that he hoped that more creators would join the virtual human image development camp, and then share it with more partners to generate greater business value.
According to relevant research reports, with the continuous maturity of virtual digital human technology, it can not only land in the blue ocean fields such as automobiles, sports, medical care, but also counterattack in the red ocean fields such as pan-entertainment.
Ten years of artificial intelligence
From this perspective, whether in the industrial field, or in the pan-entertainment or other fields, virtual human and robot are just the tip of the iceberg of AI technology. However, from a long-term perspective, today’s AI application landing is only the beginning.
For example, in iFLYTEK’s “Super Brain 2030” plan, robots and virtual humans are only the first stage of its milestones (2022-2023). In the planning, the iFLYTEK will launch a robot that can be nurtured with pet toys, bionic motion robot dogs and other software and hardware, and at the same time launch a professional digital virtual human family to play the roles of teachers and doctors.
In the second stage (2023-2025), the company will launch an adaptive walking exoskeleton robot and accompanying digital virtual human family. The elderly can walk and move normally through the exoskeleton robot. At the same time, it will launch a depression screening platform for teenagers;
In the third stage (2025 – 2030), we will finally launch a family of companion robots and autonomous learning virtual humans that understand knowledge and can learn, and fully enter the family.
From the financial report data, we can see the continuous investment of iFLYTEK in AI technology. According to Wind data, the company’s R&D expenses from 2018 to 2021 were 940 million yuan, 1.64 billion yuan, 2.22 billion yuan and 2.83 billion yuan, up 57.6%, 29.8%, 34.9% and 28% year on year respectively.
The company’s continuous investment in research and development has enhanced the company’s “sense of gain”. According to the 2022 semi-annual report previously disclosed, iFLYTEK achieved a revenue of 2.205 billion yuan in “educational products and services” in the first half of the year, an increase of 27.38% over the same period of the previous year, and a revenue of 123 million yuan in “teaching business”, an increase of 16.11% over the same period of the previous year.
In addition, iFLYTEK continued to expand its business in “smart city information engineering” and “digital government industry application”, and achieved revenue of 1.441 billion yuan and 444 billion yuan respectively in the first half of the year, with year-on-year growth of 47.93% and 66.01% respectively.
In terms of business related to auto intelligent Internet connection, in the record of investor relations activities in the previous July, the company said: “The core technical strength of AI has been widely recognized by new energy vehicle manufacturers including BYD, and BYD’s new energy passenger vehicles are all equipped with iFLYTEK products.” This to some extent means that iFLYTEK’s continuous efforts in the direction of AI have begun to earn financial returns, Start to gain public praise on customers.
However, it can be seen from the previous article that it will take at least ten years to just let robots and virtual people “fly into ordinary people’s homes”, which is far from overnight. As for the digital transformation of other industries, iFLYTEK alone is also difficult to complete. This is the meaning of iFLYTEK, the “first unit of artificial intelligence”, to deepen AI technology.
More importantly, typical scenarios, expert knowledge base and business models are displayed through the developer section for millions of developers to innovate scenarios and develop business processes to form specific industry solutions; Insisting on investing hundreds of millions of research and development funds, enabling 18 major industries such as energy, agriculture, tourism and finance through the open platform of iFLYTEK, and promoting key exchanges and cooperation through the form of industry forums, we can see the contribution of iFLYTEK in the industry ecology.
Since 2017, the iFLYTEK Global 1024 Developer Festival, which focuses on AI technology sharing, has been held for five consecutive sessions. In the sixth session of this year, iFLYTEK has adopted more than 1200 AI exhibits, focusing on the nine theme pavilions of “technology, industry, ecology, maker, enterprise, life, city, education, and health”. Participants can experience the gradual transformation of AI to the social form at a close distance.
The eve of every industrial revolution is often long and tortuous.