What Korea can learn from the DeepSeek shock

Home > Opinion > Columns

print dictionary print

What Korea can learn from the DeepSeek shock

Oh Se-jung
 
The author is a professor emeritus of physics and astronomy at Seoul National University and a former president of SNU. 
 
A small artificial intelligence (AI) firm in China sent shock waves across the world last week. DeepSeek, founded by 40-year-old Liang Wenfeng, unveiled its generative AI model, R1, which has been evaluated as being on par with OpenAI’s latest models. What is particularly astonishing is that DeepSeek operates with a research team of just around 150 people — a fraction of the work force employed by U.S. tech giants like OpenAI and Google — and developed R1 at less than one-tenth of the cost incurred by American firms. Moreover, due to U.S. export controls, China lacks access to the most advanced semiconductor chips.
 
DeepSeek’s breakthrough is raising fundamental questions about the conventional wisdom that AI advancement requires massive financial and computational resources. The stock prices of companies like Nvidia, which have thrived on this assumption, took a noticeable hit following the announcement. The paradigm of AI development may be shifting before our eyes.
 
Fifteen years ago, when a Chinese-manufactured product turned out to be unexpectedly good, people jokingly called it “a mistake from the mainland.” Back then, Chinese goods were mostly knockoffs — cheap in quality despite their polished exteriors. That is no longer the case. Today, Chinese products not only compete at the highest level but also dominate global markets in key industries. Chinese drones, for instance, have an overwhelming share of the global market, and household appliances like robotic vacuum cleaners set global trends. In the smartphone and EV sectors, China has moved beyond low-cost production and is now challenging premium global brands. The notion of “a mistake from the mainland” has long been abandoned.
 
China’s transformation has been driven by its formidable advancements in science and technology. One key indicator of a country’s technological prowess is international patent filings under the Patent Cooperation Treaty (PCT). Since surpassing the United States in 2019, China has led the world in patent applications. Huawei has consistently topped the list of corporate filers, and in 2023, four of the 10 most prolific patent applicants were Chinese companies.
 
A similar trend is evident in fundamental scientific research. China now publishes the highest number of research papers globally, and in the 2024 Nature Index — which measures the impact of academic research — the Chinese Academy of Sciences (CAS) ranked first. Seven of the top 10 research institutions in the world are now Chinese. This scientific and technological foundation is propelling China’s industries to the forefront of global competition.
 
AI is expected to shape the future of human civilization, and in this domain, China and the United States hold a commanding lead. The two countries have the largest pools of AI researchers, and over the past decade, 70 percent of all patents related to generative AI have been filed in China. Meanwhile, U.S. President Donald Trump is personally pushing the Stargate Project, a $500 billion AI initiative, demonstrating America's commitment to maintaining its lead in the sector.
 
For a country like Korea, where both research talent and funding are scarce compared to these AI superpowers, the prevailing assumption has been that it could do little more than watch from the sidelines. Yet, DeepSeek’s achievement offers a different perspective. Unlike in the past, when China merely followed in the footsteps of advanced economies, it is now pioneering new paradigms through independent innovation.
 
Skeptics question whether DeepSeek’s claims of cost efficiency are entirely accurate and whether its model truly represents a groundbreaking innovation. Nevertheless, the company’s success challenges the prevailing belief that a brute-force approach — piling on more computing power and larger research teams — is the only way forward in AI development.
 
DeepSeek’s emergence offers several crucial insights for Korea as it navigates the AI era. First, Korea must address its shortage of AI talent. DeepSeek’s researchers, including its founder Liang Wenfeng, are largely homegrown scientists who studied in China. According to a recent analysis, 47 percent of the world’s top AI researchers are of Chinese origin according JoongAng Ilbo. In contrast, Korea is experiencing a concerning talent drain. The country’s obsession with medical school admissions has exacerbated the decline of STEM fields, raising alarms about the future supply of AI professionals.
 
Second, Korea must prioritize originality over blind adherence to global trends. Competing with the United States in the semiconductor arms race is unrealistic — no country can match America’s financial muscle in securing the world’s most advanced chips. Instead, Korea should explore alternative AI development strategies that emphasize cost efficiency and novel methodologies.
 
Third, diversity fosters innovation. DeepSeek’s headquarters is not in Beijing, where China’s major tech firms are clustered, but in Hangzhou, Zhejiang. This physical separation may have insulated the company from groupthink, encouraging a broader range of perspectives and independent thinking. Korea, by contrast, suffers from extreme concentration in the capital region, stifling regional innovation hubs. The country must rethink its centralized approach to talent and technological development.
 
The world is evolving at the speed of light. Yet, Korea remains preoccupied with domestic political turmoil, unable to focus on the seismic changes happening beyond its borders. How much longer can Korea afford to remain distracted?
 
Translated using generative AI and edited by Korea JoongAng Daily staff. 
Log in to Twitter or Facebook account to connect
with the Korea JoongAng Daily
help-image Social comment?
s
lock icon

To write comments, please log in to one of the accounts.

Standards Board Policy (0/250자)