China’s Global Initiatives of Artificial Intelligence TVET
Phil S. Yang 职业教育国际研讨会 4 days ago
In recent years, I made two important decisions and put them in action. One was my participation in China’s TVET, Technical and Vocational Education and Training. With some investment, I am now International President of Jilin Cosmopolitan Vocational Technology College, Changchun, Jilin. The other was my investment and participation in artificial intelligence business. With focus on Vision AI, I am CEO of Mega AI Lab, Seoul, Korea and set up its R&D and market organizations in China, India, Canada, and now in the middle of setting up a US organization. Moreover, my AI business has expanded into a number of verticals, including logistic AI, construction AI, cosmetic AI, diagnosis AI, cybersecurity AI, and physical security AI, for which I have operated a multitude of companies; in effect, I run a group of AI companies. Right now from the background behind me, you may see a variety of our AI technology which can be experienced by any visitor to our building.
This opportunity provided for me by <2021 Belt and Road International Conference on TVET> is not only a great honor to me but also a great chance for me to combine what I did in the past, what I do now, and what I will do in the future. In particular, I will take advantage of the conference as an opportunity to have collaboration with more TVET institutions in China. And I would like to express my gratitude for speech invitation to the leaders and staffs of the conference and its participants.
Belt and Road Initiatives for TVET
Today’s conference is part of efforts made to integrate the advance of the digital economy, the improvement of technical and vocational talent, and the global cooperation of TVET on the basis of China’s initiatives of Belt and Road. In 2017, I was lucky enough to be invited by the General Executive Bureau of Silk-Road Industry and Finance International Alliance to write my opinion about China’s Initiatives of Belt and Road. In order to understand Belt and Road in depth, I decided to do it with the assistance of my students at Tsinghua University in Beijing. When it was published, I was surprised to find my name along with Premier Li Keqiang and Mr. Chen Yuan and others in the list of contents. Indeed, it was a great honor to me.
China’s Belt and Road initiatives are to rise from the “World Dream,” Shijiemeng, which was, in turn, extended from the “Chinese Dream”, Zhongguomeng. It is not simply the Chinese people who pursue and enjoy the fruits of such a development, but the people all over the world are pursuing and sharing those of the Initiatives. To achieve such a shared dream, the Initiatives should go beyond a simple concept of economic exchange; rather, they should be realized through comprehensive exchanges, such as cultural and educational exchange, in particular the exchange and cooperation of TVET in the Belt and Road network.
Even though China witnessed its long history of TVET, it waited a while for its institutionalization. Today, the government pays more serious attention to it. Not only in China but also in the whole world, including Korea, jobs have becomea critical issue of society; the education and training for jobs have been especially important to young people. In the case of China, two challenges emerge to TVET. One is quality over quantity. The other is internationalization.
In this context, teaching and training artificial intelligence become crucial to the quality of TVET; international exchange and cooperation in the Belt and Road network are not only important to the Belt and Road Initiatives themselves but also to upgrading the quality of each TVET institution in China. Now let’s turn to what artificial intelligence in general, deep learning in particular is and how it is immersed into the pedagogy of TVET.
Artificial Intelligence and Deep Learning
In a near future, every company will be an artificial intelligence company. When? I think it will be within 10 years in China, much faster than in other countries. What is artificial intelligence? I would like to tell you about the history of artificial intelligence, that is., AI. In 1956, a number of researchers and professors got together at an American university called Dartmouth College, one of the prestigious Ivy League universities, to which my daughter went twenty years ago. They discussed the possibility that machine could think like human beings, which was called as artificial, not real, intelligence. Since then, there have been ups and downs of that technology.
In history of AI, the year of 2012 was important as much as that of 1956. Two great events occurred almost simultaneously. On the software side, what is called deep learning was innovated by a number of Canadian researchers; they coded computer programs working like human brain. Like the neural network of human brain, computers learn and think by themselves on the basis of data. That software program is called algorithm.
On the other hand, new process chips were manufactured to deal with graphic data in a massive volume, which is called GPU, graphic process unit. This innovation entailed a large capacity of computing power which in turn requires that of electricity as well as a large amount of capital investment. Around the year of 2012, these new trends merged together and brought about a new technology of deep learning, a subset of machine learning which forms part of artificial intelligence.
In 2016, the world was surprised to witness AI beat Lee Sedol in a five-game match of Go. Although it lost to Lee Sedol in the fourth game, Lee resigned in the final game, giving a final score of four games to one in favor of AlphaGo, an AI computer program. The next year, AlphaGo beat Ke Jie, the number one ranked player in the world at the time, in a three-game match. Since then, the self-taught AlphaGo Zero achieved a 100–0 victory against the early competitive version of AlphaGo, and its successor AlphaZero is currently perceived as the world’s top player in Go.
It was a shocking event. However, the Chinese leaders and businessmen did not just watch it but they took immediate action. In 2017, the government began to encourage venture capitalists to make investment in AI technology in order to achieve the new goal of making China as Number One AI Country in the world. With more data and computing power, Chinese engineers and businessmen took advantage of the recent advance of deep learning technology, which mimic human brain’s neural network.
Like the human cognitive system, deep learning is composed of voice AI, vision AI, and language AI, and recently even smelling AI. In general, people call voice AI as voice bot or chatbot which simulates human conversation through voice commands. Vision AI is a field of computer science that trains computers to replicate the human vision system, that is, see and understand what is. It is still called by many people, including experts, as computer vision, which is wrong; we should distinguish data-driven vision Ai from rule-based computer vision. Finally language AI is called as natural language processing, that is NLP. Even before the emergence of deep learning in 2012, efforts were made to develop voice AI like chatbots, which failed to monetize to a great extent.
While computer vision, which emerged in the late 1960s, was not widely popular, especially commercially, it was vision AI that made money due to its diverse applications, such as autonomous driving, security, safety, agricultural drone, and so many applications. Because Chinese people have, indeed for a long period of time, been engaged in digital life, especially through the use of smartphones, China had a competitive advantage of visual digital data and vision AI technology against any other country, including United States. This means the great potential for China to be Number One AI country in the world; as a matter of fact, in some fields of AI, such as vision AI, China has already surpassed America which remains Number One in many areas of AI yet.
AI Education and Training for TVET Students
In my opinion, Shenzhen will become, or has already become, one of the major centers for AI technology. Interestingly enough, the city has everything from sensor suppliers, injection-mold engineers, small-batch electronics factories and so on. Some people now do not say “made in China” but rather “made in Shenzhen.” More importantly, it has the unparalleled flexibility of the supply chains and the abundant pipelines of skilled field engineers who can make prototypes of new devices and build them at scale.
To me, it is the city’s culture that makes it a true crown for hardware and a new Chinese Silicon Valley for software. For example, a week of Shenzhen amounts to a month of America. While Zhongguancun in Beijing is now competing with Silicon Valley in the United States, Shenzhen will come to compete with it in a near future due to the nature of AI technology in general, deep learning in particular, which entail both software and hardware power. In this context, it is indeed meaningful that today’s conference is held in Shenzhen, especially with my speech topic of AI education and training for TVET students.
As Li Kaifu, one of the leading AI experts points out, it is very unlikely to see another round of technology breakthrough in the field of deep learning, which means that there exist no chances for deep learning to have such a major achievement or advance in a near future as that of 2012. In effect, the quality of AI algorithms for deep learning depends on more a volume of data than a novelty of proprietary research. In this context, more middle-level engineers are needed than elite engineers with Ph.D. or Master’s degree; with strong computing power and humongous Big data, those who graduate from TVET schools are able to encode more powerful algorithms than elite AI experts, if the latter cannot afford to possess computer power and data volume as much as the former.
The unique situation of current AI technology leads us to find a way how TVET nurtures AI talent. First of all, there are great demands for middle-level AI talent not only in China but also in other countries; the more AI deployment, the more AI field engineers. As pointed out earlier, it is not difficult for those who finish TVET to create novel AI algorithms which have commercial value. For the students of TVET, learning AI-legacy hybrid technology is more important and useful than learning its state-of-the-art knowledge and skills, especially for their employment. Here, a legacy technology is defined as an old technology or application program that is yet still in use but also is paving the way for the standards that will follow it; in particular, in computing, a legacy system refers to software or hardware that has been superseded but is difficult to replace because of its wide use. On the other hand, a hybrid is a mixture of two different things, resulting in something that has a little bit of both.
In the real world, such disruptive technology as deep learning AI is not rapidly adopted by companies because of its costs and human bottleneck. A combination of AI and traditional IT is a natural course of AI adoption for most of the companies in the real world. As a result, much more demands have emerged for the AI-legacy hybrid technology than for the state-of-the-art AI technology which is considered as important in the research laboratory and universities. Finally, as pointed out at the beginning, every company will become an AI company whatever industry it belongs to. That means more and more jobs will be available for TVET students who learn AI technology.
TVET Schools' Challenge for AI Education
Of course, the TVET institutes have a challenge to overcome. Above all, teachers who are able to teach AI are not enough in quantity for TVET. This is also true for universities, including Tsinghua University in Beijing. At today’s conference, I would like to suggest that a collective effort be made to develop teaching curriculum and material for the AI education of TVET students. Now that due to the corona virus, virtual education becomes popular, video and virtual material become necessary and cheap for the AI education of TVET in the form of collective teaching. Once again, I would emphasize that the content of TVET education should focus on the AI-legacy hybrid technology.
The students of TVET will be able to have the easy access to high-paid employment by taking advantage of their AI knowledge and skills. As companies need more middle-level AI engineers who are not necessarily either Ph.D. or graduate of four-years colleges, the students of TVET have more chances to have employment especially due to their familiarity with the domain knowledge and experience in a specific industrial field. In this context, combined with the tradition of field-focused learning, the TVET students develop their AI knowledge and skills that the real world and their employers desperately need. From my observation as university professors in Korea, China, and the US, the knowledge and skills learned from the four-year universities tend to have a distance from the real world, which is not a matter of good or bad things, though. It is just the division of labor between university education and TVET.
In the meantime, the TVET institutes should be ready for teaching how to organize and manage start-ups. The digital economy facilitated by AI technology will require more startups than the traditional IT economy. More applications are needed to apply AI technology to the real world. As I pointed out earlier, the proprietary algorithms which are responsible for running the main engine of deep learning will remain the same for a while; in the field of deep learning which possesses a lot of application solutions with commercial value, no technology breakthrough will occur in a near future. Now it is a time of big data and computing power that facilitate the development of AI technology in general, deep learning in particular, which will bring more jobs and business opportunities than any other subset of AI technology. As a result, this will bring about more business opportunities for startups.
One more thing should be emphasized right now. Open sources for the business development of AI technology are widely and deeply available, especially in China. This is accounted for by the government policies of “Mass Entrepreneurship,” dazhong-chuangye, which encourage the ordinary people, from farmers in the village to merchants in the city, to do their own business by taking advantage of new technology, including AI. To be sure, a variety of open sources for AI technology helps the TVET students with their own entrepreneurship. In order to help the students with startups, the school needs to provide incubation experience for them. For instance, several school get together to build and run their joint incubation centers with the cooperation from AI companies; while taking class from their own schools or the joint classes, the students and teachers together set up their own companies in the incubation centers.
In my opinion, the idea of collective efforts for TVET’s AI education may be extended to a global scale. That is also a way to achieve China’s Initiatives of Belt and Road in the realm of TVET. To be sure, the global cooperation of TVET may start with the collective efforts to teach AI technology in the Belt and Road network.
AI Education for the Belt and Road Initiative
Now China forms part of AI G2 along with the United States, which means that China and United States are leading the AI technology of the world. The country has competitive edge over big data; due to its wealth, its computing power is also competitive.
By taking advantage of those superior elements, middle-level engineers can have well-paid jobs and set up their own enterprise in China; the TVET institutes are, in effect, responsible for nurturing them. Inside of China, the TVET institutes make collective efforts to nurture them in order to fill up the gap between the future goals and the current capacities.
Because local data is more important than local algorithms, AI requires a higher degree of localization for globalization than internet services. Some algorithmic training can be transferred between different user bases, but it is not possible to substitute the data of another country for the local ones of one country. Simply to put it, this is just because data are not uniform but local. With regard to AI talent, the extent of localization is more demanding; as a result, more local engineers should be provided for each country. This is to lead us to realize the importance of global exchange and cooperation for the AI education and training of the TVET students.
In order to put the agenda of global cooperation in the context of the Belt and Road Initiatives, that logic of idea may be simply extended to outside of China. Collective efforts across national boundaries will be made to develop and share the teaching curriculum and materials for the AI education and training of the TVET students. Now that it has abundant resources of AI talent and technology, China will play a leading role in the TEVET network of AI education and training. This will, to be sure, contribute to the advance of China’s Initiatives of Belt and Road, which will, in turn, facilitate the development of TVET in China.
Through the global cooperation of TVET in the field of AI technology, other countries and their people, including those from Asian and African countries, will come to consider it as more friendly and positive. Thanks to the AI education and training in the Belt and Road network, the young generations of the world will have better life and higher income and as well more opportunity to understand each other regardless of their color of skin and language of speech. To be sure, the AI education and training for TVET students will not only improve the quality of China’s TVET system but also promote the globalization of it, which are its two great challenges today.
It is high time for us to work together not only between TVET institutions in China but also between those from different countries. Let the Belt and Road Initiative guide us how to do it across the national boundaries. And this will lead the Chinese institutions of TVET education and training to reach the next level of development and to achieve their own globalization agenda. Let us not wait but do it now and do it together.
Conclusion
Only within two years, I achieved a rapid development of AI technology more for use application than for proprietary research. Now I came to have a group of AI companies with focus on Vision AI. I consider such a success as the result of my globalized efforts to take advantage of hardware and software pipelines from all over the world.
Even though I offered a high amount of salary for them, I failed to hire a lot of Ph.D. engineers as many as I originally looked forward to having. This, however, did not end up with total failure but rather served as a new opportunity. Instead of heavy investment in AI talent, I made huge investment in hardware, including our own super computer with two petaflops which run two quadrillions of data per second and build the global network of hardware and software pipelines on the basis of our own global R&D and marketing organizations.
Two things are learned from my own experience; problem-solving capacities and globalization. The real commercial values of AI technology came from the capacities of problem-finding and solution-execution, both of which TVET institutions have traditionally placed great emphasis on. In effect, globalization helped me to resolve the lack of AI talents. To be sure, these two elements will also serve as the great contribution factors for China’s initiatives of AI education and training for TVET students at a global scale. On the basis of quality TVEST education and training, Shenzhen will become a global AI hub different from Beijing in China and Silicon Valley in US; I wish my experience and insight to help Shenzhen with such a path.
Thank you very much!
Reference
Christopher H. Lim, Vincent Mack Zhi Wei, “‘Chinese Dream,’ Global Ambition: Beijing’s Belt and Road Initiative,” Global Asia, Vol.13 No.3 (September 2018).
Iansiti, Marco & Lakhani, Karim R. 2020. Competing in the Age of AI: Strategy and Leadership When Algorithms and Networks Run the World. Harvard Business Review Press.Boston, MA.
Lee, Kai-fu (2018) AI Super Powers: China, Silicon Valley, and New Old Order. New York: Houghton Mifflin Harout Publishing Co.
MBN China Research Team (2018) What Makes China Number One. Seoul: Mail Business Newspaper Press.
Sun Jin & Sun Ning (2015) Employment in China: Modernization of Vocational Education. Tsinghua University Press.
Yang, Dongha (2020) High-Technology and Intellectual Property Rights from Marxist Perspective:
Case Study of Huawei and ZTE. MS Thesis of Tsinghua University, Research Institute of Marxism.
Yang, Phil Seung (2017) “One-Belt and One-Road Perceived from Chinese Dream to Global Dream” No. 5. Newsletter of Silk-road Industry and Finance International Alliance.
Comments