A few months ago, I participated as a panelist in a seminar titled "The Future Changes Brought About by Generative Artificial Intelligence (AI) and the Response Direction of Our Companies" which was organized by the Korea Chamber of Commerce and Industry in collaboration with a non-profit organization called Tech & Trade Forum.
Various technologies related to the 4th Industrial Revolution are being contested between the USA and China. Among them, perhaps the fiercest battle is being waged in the field of AI. Undoubtedly, AI is poised to bring about the most significant changes and impacts to global politics, economies, industries, and people's lives. Currently, the US and China are engaged in a major strategic game in fields closely related to AI. While AI technology itself is crucial, semiconductors and big data are mentioned as areas closely tied to AI. To run the enormous algorithms and computations of AI, regular computers or processors aren't enough. Instead, semiconductors that can handle and process vast data tailored for AI are needed. NVIDIA, a prominent AI semiconductor company, has seen its value skyrocket over the past five years. South Korea, proud of its semiconductor powerhouses like Samsung Electronics and SK Hynix, still falls short in the AI semiconductor field that will determine future global dominance. China has yet to make a significant mark in the semiconductor sector but is investing heavily to advance its capabilities. With the recent indications by President Xi Jinping of China regarding Taiwan, there's growing tension globally. One major reason for China's interest in Taiwan could be to acquire TSMC, Taiwan's dominant semiconductor company. If China controls TSMC's technology and infrastructure, it could deal a significant blow to US semiconductor capabilities.
The second field of contention is big data. Most internet and mobile users worldwide have apps like Google, YouTube, Facebook, Instagram, Amazon, and Twitter. These platforms generate vast amounts of big data. However, China remains inaccessible to these companies. Instead, China uses its platforms like Baidu, WeChat, and Alibaba based on its massive population. These platforms, despite being China-centric, generate vast amounts of big data which helps enhance China's AI capabilities. Both the US and China, two giants, are now solidifying their walls in the AI field, prohibiting any collaboration or tech-sharing, especially in the semiconductor and big data domains. AI technology will inevitably be the determining factor in the future of global economics, trade, and security. Interestingly, with the rise of generative AI, as exemplified by ChatGPT, both China and the US continue their fenced competition.
It raises concerns and hopes for smaller countries like South Korea. How can they survive and thrive in this massive generative AI market dominated by these two giants?
Many outstanding individuals participated in the forum that day. Notably, Dr. Ha Jung-woo, the head of Naver's AI Research Institute, stood out. As the leader of AI at Naver, South Korea's top IT and big data company, I was curious about him. Having met him, he indeed lived up to the expectations. Before the forum, we had a tea time of about 30 minutes, during which he shared some hopeful insights. Many might wonder how companies like Naver from South Korea will compete and profit against behemoths like ChatGPT and Google Bard, but Dr. Ha's answer was simple: "We have our market and a global niche market." Only three countries have made significant progress in large-scale language model AI: the USA, China, and South Korea. It's not surprising when you consider that not many countries have powerful search engines and communication platforms like South Korea's Naver and KakaoTalk. In contrast, countries like the UK, France, Germany, Italy, Canada, and Australia predominantly use platforms like Google and Facebook. What was even more pleasant to learn was about Japan. Although I knew Naver's Line was widely used in Japan, like KakaoTalk in Korea, I was unaware that Naver also provides and operates the search engine for Yahoo Japan.
Another good story I heard from him is that there is a niche market for Naver to create and provide ultra-large generative AI models globally for each country. Countries like Saudi Arabia in the Middle East, the United Arab Emirates, or nations in the Russian sphere, which resist U.S. dominance, many countries in Central and South America are not friendly towards the U.S. These nations hesitate to become dependent on the U.S. or have their country's data and information drained to the U.S. Then there are third-world countries like India or Southeast Asian nations which have unique language models not in English or Chinese. Having developed a unique language like Korean for several years and having a somewhat verified generative language model, they can easily develop models for these countries' unique languages.
Of course, we can't come close to the vast capital, technology, and already established big data training of American companies like MS or Google. Still, if we can build and operate models based on each country's unique language quickly and cheaply using our own technology, our country will have a significant business opportunity in the ultra-large generative language AI model. Domestically, Naver has long had a platform recognizing and processing the CLOVA language and offering chat and response functions. They plan to expand this further, offering the Hyper CLOVA concept. They are said to collaborate with Samsung Electronics in the learning process of this AI model, with Samsung providing powerful AI semiconductors, ensuring the costs and energy of the training. They plan to launch Hyper CLOVA X around July, enhancing the standard language model Hyper CLOVA by integrating big data from specialized fields, like architecture, law, medicine, arts, etc. When these specialized data are further trained, the AI becomes more expert in those areas, providing more detailed AI services.
They're also planning a project to build a 'Digital Platform Government,' opening up all public documents and data so that AI can read and process them. This will improve the quality of public services by providing information and services more easily through enhanced call centers or chat services, heralding a new era of public generative AI.
If Naver takes the lead in our country in building such various field projects or platforms and creates success stories, this will present significant business opportunities, especially in countries hesitant about U.S. or Chinese influence, including neighboring Japan. China has its barriers and a huge domestic market, so their interest in such niche markets remains low. Also, many countries, while not friendly with the U.S., don't have good relations with China either. In such unique ultra-large generative AI language model markets, Korea's position will undoubtedly stand out. Behind this, we believe the globally popular and loved K-POP, K-Drama, K-Movies, and other K-Cultures will firmly support it. Seeing the young Dr. Ha Jung Woo explaining Naver's ambitions with the ultra-large generative AI without hesitation made me feel proud and supportive. However, as someone who received a Ph.D. in Computer Science with AI 26 years ago, I couldn't help but feel that if I were 20 years younger, things might be different.
Professor Jang Jungho
Comments