*Editor’s note: K-VIBE invites experts from various K-culture sectors to share their extraordinary discovery about the Korean culture. 

 

Matthew Lim's AI Innovation Story: The Emergency Era of Securing AI Talent

 

By Matthew Lim, AI expert and director of the Korean Association of AI Management (Former Head of Digital Strategy Research at Shinhan DS)

 

 

On May 31, I was at a loss for words after reading about the “National AI Talent Development Project” announced by the Ministry of Science and ICT.

 

The plan itself—attracting around 400 postdoctoral researchers from overseas with a total budget of 30 billion KRW—sounded fine. The problem came next.

 

“The base annual salary will be guaranteed at 90 million KRW. This is about 1.8 times the average salary of postdoctoral researchers in Korea, which is around 50 million KRW. We will also work with industry to provide additional incentives.”

 

Seeing the government promote this as a bold benefit reminded me of a scene from the movie "Moneyball". What if Billy Beane, the general manager of the Oakland Athletics, had said, “We’re offering $100,000, so the best hitter in the major leagues will come”?

 

It would have been a joke across the league.

 

The real issue lies in the reality of overseas talent.

 

Let’s look at the actual conditions of the talent we’re trying to recruit from abroad. It’s not the time to be comparing salaries to domestic postdocs.

 

In the U.S., postdoctoral researchers in AI and related fields at universities or public institutions typically earn between $60,000 and $75,000 (around 80–100 million KRW) as starting salaries. With experience, this can rise to $65,000–$75,000.

 

But the industry is another story. Postdocs in companies specializing in AI or computer science receive between $90,000 and $130,000 (roughly 120–180 million KRW).

 

In Silicon Valley, the average salary for AI research scientists at Meta is $330,000 (around 470 million KRW). Even at a typical AI startup, PhD holders are said to receive a minimum of 200 million KRW annually.

 

Can we realistically expect such individuals to come to Korea for 90 million KRW?

 

That’s like offering a minor league salary to a Major League All-Star.

 

What’s more concerning is that, according to the Korea Software Industry Association, the average salary for AI developers in Korea is about 85 million KRW, and according to Wanted, the average salary of successful applicants for AI-related positions is 77.7 million KRW. So the government’s 90 million KRW offer isn’t particularly competitive even by domestic standards.

 

Even the figure of 400 is misguided.

 

So here’s the fundamental question: is 400 a meaningful number?

 

Rather than distributing 30 billion KRW across 400 people, wouldn’t it be better to focus on half that number—say 200—and provide them with genuinely competitive packages?

 

If the average U.S. AI engineer earns $180,000–$320,000 (about 250–450 million KRW), then we should at least offer packages worth 200–300 million KRW to be even halfway competitive.

 

That would make 100–150 people a far more realistic target.

 

Now is the time for "jimo-hyeonhyeon (智謀賢顯)"—for wisdom and strategic thinking to shine. Quality over quantity.

 

In the world of AI, 10 world-class researchers can drive more innovation than 400 average ones.

 

But there’s an even more critical issue than money: deep-rooted structural problems.

 

◇ A System That Doesn’t Support Long-Term Research

 

One of Korea’s biggest problems is the excessive pressure for short-term results. Whether in government-funded projects or corporate R&D, demands like “deliver a demo in three months” or “make it commercially viable within a year” are common.

 

According to the Stanford AI Index 2024, training GPT-4 cost about $78 million, and Google’s Gemini Ultra cost $191 million in computing resources. Expecting quick results from such research-intensive efforts is simply unrealistic.

 

A more fundamental issue is Korea’s rigid retirement system. University professors must retire at 65. After decades of dedicated research, many are forced out due to this age limit.

 

Even if they become emeritus professors, they’re often restricted to teaching one course per semester for only three years. This causes a sudden and tragic loss of accumulated knowledge and expertise.

 

It doesn’t compare to the U.S. tenure system. In America, tenured professors can continue working until 80 if they choose. It’s common to see a 73-year-old PhD advisor saying they plan to retire “in 3–4 years” while still actively conducting research.

 

The tenure system offers researchers the autonomy to pursue bold and risky ideas instead of playing it safe. This kind of long-term stability is essential in fields like AI, where both foundational and applied research take significant time. We need environments where experienced researchers can focus without worrying about arbitrary retirement rules.

 

So what should be done?

 

We must benchmark the U.S. tenure system and create a domestic policy that ensures long-term research stability for top researchers. The government and private sector must guarantee AI researchers a minimum of 3–5 years of free, independent research.

 

Programs like Google’s 20% time or 3M’s 15% time—where researchers pursue their own projects—should be formally implemented.

 

◇ Rigid Hierarchical Culture

 

The AI research culture in Silicon Valley is astonishing. It’s common to see a PhD student in their 20s debate ideas on equal footing with a senior researcher in their 50s—and sometimes even lead with more radical ideas. In this culture, the quality of ideas takes precedence over age or rank.

 

What about Korea’s research culture?

 

Here, the expectation to “just do what your senior says” still runs deep. In a fast-moving field like AI, such rigidity is fatal. Innovative ideas from young researchers are often buried under hierarchy.

 

To change this, we need to encourage horizontal communication within research labs and implement systems that ensure creative ideas from younger researchers are not suppressed. Korea should also expand joint research programs with global institutions like Stanford, MIT, and Oxford, especially in AI-converging fields like semiconductors, batteries, and robotics—areas where Korea already has a competitive edge.

 

Even if we manage to recruit talented researchers, it means little without a proper research environment. According to the Stanford AI Index 2025, the scale of AI models continues to grow rapidly, with training compute doubling every five months.

 

It's only natural that researchers wonder, “Will I still be able to conduct world-class research if I move to Korea?” Korea must ensure world-class standards in computing resources, data accessibility, and research facilities. Without that, even the most attractive compensation won’t bring truly capable researchers.

 

The very notion of attracting top AI talent with a 90 million won salary shows a fundamental misunderstanding of the scale and pace of the AI revolution.

 

Fortunately, on June 15, the Yoon Suk-yeol administration appointed Ha Jung-woo, head of Naver Cloud's AI Innovation Center, as the first Chief of AI Future Planning. Ha, who oversaw the development of Naver’s HyperCLOVA X, is regarded as an expert in both theory and practice, as well as policy. He has emphasized that the AI chief should go beyond mere administrative tasks and establish a practical roadmap.

 

As someone who has worked directly with AI talent, Ha understands what they need and in what environment they thrive. We can hope that, under his leadership, the structural issues outlined above will be addressed.

 

Rather than offering unrealistic deals like 90 million KRW, we need to propose attractive, practical packages that make Korea a real option for top-tier global talent.

 

To become a true AI powerhouse, budget injections alone won’t suffice. We must transform our research culture, societal systems, and mindsets.

 

Let’s not be obsessed with the number 400. We need a strategy focused on fostering a select few with real competitive edge.

 

It’s not too late—but we must face reality and begin fundamental change. If not, we’ll be relegated to the sidelines in the age of AI.

 

 

[ⓒ K-VIBE. 무단전재-재배포 금지]