Prof. Kyoung Mu Lee, Urgent discussion on ‘AI National Strategy’ with field experts (, 20191101)

2020-01-30l Hit 813

At the Seoul Central Post Office, Jung-gu, Seoul on October 31st, Kiyoung Choi (left), Minister of Science and ICT, conversed with SNU ECE Professor Kyoung Mu Lee (middle) and Yejin Choi, Professor in the School of Computer Science and Engineering, University of Washington, on the future path of the ‘national strategy on AI’. On this day, Minister Choi emphasized that for professors of AI related departments, the government would relax regulations that prohibited them from simultaneously holding positions in private corporations to promote the interaction between their research capabilities and the companies’ on-site experience.


On the 31st, Minister of Science and ICT Kiyoung Choi met a Dong-A Ilbo reporter and stated, “I will loosen the restrictions that block professors from concurrently holding positions in private companies for those in AI related departments, thus encouraging the fusion of their research abilities and companies’ field experience.” He announced that he was discussing all measures including the legislation procedure with the Ministry of Education. He added, “On a medium to long term basis, the elementary and middle school curriculum will also be modified to raise AI talents.”


On the 31st at the Seoul Central Post Office, Minister of Science and ICT Kiyoung Choi said, “Although it is true that Korea’s AI industry is two years behind compared to leading countries such as the U.S. or China, there are still plenty of opportunities left.” On this day, Minister Choi had a discussion with SNU ECE Professor Kyoung Mu Lee and Professor Yejin Choi from the School of Computer Science and Engineering of the University of Washington who represented the AI industry of Korea and the U.S, respectively. He also disclosed the framework of the ‘national strategies on AI’ that would be announced near the end of the year.


The national strategies related to AI included sensational drafts including repealing prohibition of professors in AI related departments from holding offices in private enterprises at the same time. Currently, professors are not permitted to hold other positions for profit according to the State Public Officials Act and the Private School Act. Certain exceptions can be made, such as allowing professors to be CEOs of venture businesses according to the Act on Special Measures for the Promotion of Venture Businesses or allowing professors of private colleges to concurrently hold profit related positions with the permission of the university president.


Professor Choi said, “Even though I am a professor, I am also working as a senior research manager at the Allen Institute for Artificial Intelligence, a private research institute. The U.S., which is at the frontier of the AI industry, is currently allowing professors to hold other offices concurrently.” She emphasized that it is necessary to quickly absorb things that can be learned from overseas. Following are the questions and the responses.

Q: Based on your assessment of the current state of Korea’s AI industry, at which level have we arrived? What are the strengths of Korea and in which manner do you think Korea’s AI should proceed?

▽Minister Kiyoung Choi=Currently, the U.S. is ranked first in the AI field. According to an analysis by a professional agency, Korea is two years behind the U.S. However, I believe that the gap can be overcome depending on future proceedings. Korea has a solid ground in hardware, especially semiconductors, which constitute the foundation of AI. With advancements in the IT (Information Technology) industry, leading global IT companies were switched from Intel (microprocessors) to ARM (imbedded devices) and NVIDIA (GPU). We should also follow this change of current. We aim to utilize our advantage in hardware and encourage the fusion of semiconductors and AI industry.


▽Professor Kyoung Mu Lee=We cannot achieve everything no matter what we do. At this point, we must make choices and concentrate on certain fields with our limited time and human resources. As the minister stated, we must find and solidify Korea’s unique position in the market that utilizes our strength in hardware and combines that with software.


Q: What will the ‘AI National Strategy’ that will be announced on December contain? How will the 1.7 trillion won budget for related fields be allocated?


▽Minister Kiyoung Choi=The announcement will concern four fields including education, industry, government, and infrastructure. In the education field, through discussion with the Ministry of Education, the elementary and middle school curriculum will be revised to become more AI friendly. For universities, based on collaboration with the Ministry of Education, we are proceeding with the legislation of laws that allow professors in AI related departments to hold positions in other offices. In the industry, we will search for methods to further apply AI in various fields including manufacturing and medicine. The government will also be digitalized. As pertaining to infrastructure, we will come up with measures on key resources including data and will invest in foundational technology of AI chips. Amongst ourselves, we call such new born industries ‘DNA’. Each letter stands for data (D), network (N), and AI (A). Approximately 1.06 trillion won will be invested in encouraging data production and circulation and the construction and utilization of an AI ecosystem. Furthermore, for the network field, the remaining 640 billion won will be invested in the public and private investment of 5G mobile communication and in the establishment of an industrial foundation.


Q: In Korea especially, as well as other countries, the issue of the scarcity of talents in the AI field remains unanswered. Do you believe that these methods will resolve this issue?


▽Minister Choi=One of the key issues covered by the comprehensive measures to be announced within this year is about raising talents. Academia alone cannot resolve this problem. Because the U.S. allows professors to concurrently hold other jobs, companies such as Google and Facebook offer positions to many researchers in college. Joint projects are also more common. Currently, Korean professors can conduct projects in private companies under restrictions such as by taking a sabbatical leave. Even if they undertake a project, it is difficult for them to have direct access to the company’s information. In AI, the collaboration between academia and industry is crucial. This is why we hope to carry forth a plan to relax regulations prohibiting professors of AI related departments from holding another job.


▽Professor Choi=If professors are allowed to work as members of private companies, the synergistic effects are significant. Researchers are frequently confronted with the issue of how much data and how many GPUs are available when conducting research, and this cannot be simply resolved with funding alone. From the perspective of companies, if you recruit a professor, the students from the lab also participate in the work. The students can conduct research with the actual data provided by companies and this can pose as an opportunity for them to make a contribution to the industry in the future.


Q: A flourishing AI industry requires the unrestricted use of big data that is to be analyzed with AI. However, there are too many regulations including the Personal Information Protection Act that pose obstacles for the use of data.


▽Minister Choi=A while ago, when I met Hae-chan Lee, leader of the Democratic Party of Korea, he said that he definitely would make revisions to the three acts concerning data (Personal Information Protection Act, Act on Promotion of Information and Communications Network Utilization and Information Protection, etc., and Credit Information Use and Protection Act). So, I told him, “Then I will no longer worry myself over the Personal Information Protection Act.” For example, in Europe, the General Data Protection Regulation (GDPR), a comprehensive regulation, was created, and with this single regulation, it has become possible to provide the foundations for the protection and use of personal information as well as to offer a guideline for the AI industry. As can be observed from this example, not only is a revision of the Personal Information Protection Act directly integrated with obtaining data that becomes the ground for development in the AI industries but also may later prove to be a steppingstone for collaboration with leading countries in data.


▽Professor Choi=In the U.S., there are several cases in which Google, Facebook, or any other company that leads AI research manage research projects on a single topic on two different levels, with research based on internal data and data that will be open-sourced. In the case of the latter, not only the obtained data but also the code and analytical model used for the research are shared and this is significantly contributing to academia. Facebook, which was lagging behind Google in the AI field in the early stages, has recently soared to the top thanks to open research.


Q: Recently, there has been a rising necessity of growing independent of other nations when it comes to the material, components, and equipment industry. There is now a general consensus that there is ‘need for long term investment and research’ in Korea. In the long term, what do you think is the correct approach for long term research and investment?


▽Minister Choi=Actually, deep learning and other research fields that are attracting attention are not novel concepts. In the past, ‘neural networks’ enjoyed the limelight in science but soon returned to the shadows. Interest in deep learning resurfaced thanks to the people centered in Canada who had not relented and had further advanced their research in that field. Through this example, one can infer that persistent investment in the fundamentals, along with creativity and brilliant ideas, is important.


▽Professor Lee=An analysis of the one thousand papers registered for this year’s International Conference on Computer Vision, the most notable conference in the AI and pattern recognition field, revealed that about 330 were from China and 310 were from the U.S., adding up to account for 64% of the work. Next up was Germany(47) and Korea(44). There were 2900 Koreans among the 7500 people registered to participate in this year’s conference. This figure evidences the sheer amount of interest Koreans have in this field. This will serve as the foundation for Korea’s AI in the future.


▽Professor Choi=Recently, if you look into the AI academia overseas, noteworthy research have been presented by unremarkable organizations that are not included in top ranking schools or conferences. What this signifies is that because this still remains a novel field, if you have creativity and the will to challenge yourself, anyone can conduct valuable research.


Q: The government’s role in the development of Korea’s AI industry is crucial at this moment. If you were to offer your opinion to the Korean government as experts in Korea and the U.S.’s AI field…


▽Professor Choi=The U.S. is keeping a watchful eye on China’s rapid process. The government has gone as far as to consider whether ‘it is an AI research field that China has not begun and a field in which the U.S. must take lead’ when choosing an AI project for investment. A research project my team is looking into with the support of the United States Department of Defense was selected based on such judgment. Perhaps the Korean government should support research in fields that have been left untouched by competitors and whose long-term characteristic makes it difficult for private enterprises to make an attempt.


▽Professor Lee=Another problem is that there is no market established for startup businesses. About 70% of domestic startups choose to sell rather than list their companies to redeem invested capital. Although the major companies in Korea maintain that they invest in AI, they are reluctant to offer large sums for AI startups of technologies. The fact that the culture is not M&A (mergers and acquisitions) friendly is also taking its toll. It is necessary to establish a virtuous cycle where various AI startups are born, developed, and sold to major companies.



Translated by: Jee Hyun Lee, English Editor of Department of Electrical and Computer Engineering,