“After my appointment, when I visited a research institute, they were demonstrating something that Samsung had done 3 to 4 years ago. I couldn’t help but ask, ‘Why are they doing this?’
Kim Hyun-seok, former president of Samsung Electronics and current head of the Ministry of Trade, Industry and Energy's R&D Strategy Planning Team, shared an absurd experience he encountered while visiting major domestic research institutes during an interview held on the 20th.
The research institutes were conducting R&D on similar topics and content. It is reported that some research projects had already been completed by the corporations.
He raised his voice, saying, “Someone needs to organize this.” He added, “We need to clearly divide the areas where each can excel and manage the division of labor well. We need to work closely with the corporations.”
Kim noted that in a situation where global research institutes, including those in the U.S., are attracting talent with high salaries, enhancing the foundational capabilities of domestic R&D personnel and building teamwork is the top priority assigned to him.
He also said, “We need to change the practice of dividing R&D projects finely to distribute them among various places,” emphasizing that the structure should shift from one person managing multiple small projects to several researchers working on projects worth hundreds of millions or billions.
Regarding Korea's AI capabilities, he assessed, “We are still at a beginner level,” pointing out that for AI development, the priority should be on ‘industry-specific AI’ (domain-specific generative AI) rather than public LLM (large language models). He mentioned, “Although the currently discussed sovereign AI may have significance as a public good, it will be difficult for it to serve as a strategy leading the entire industry,” and said, “We need to establish a concrete roadmap to create practical AI tailored to each industry, not just universal models.”
Kim stated, “If we do nothing now, we could truly be miserable,” and added, “I celebrated too early; when I opened my eyes, I realized we’re in a situation where even ‘fast follow’ has become difficult.”
Next is the question and answer session.
―Let’s talk about regulations.
“President Lee Jae-myung also said he would drastically ease regulations, but what needs immediate attention is data regulation. Our country makes it really difficult to utilize data. The government holds a vast amount of data, yet it does not open it up. For developers, data is crucial.”
―Are you suggesting that data openness should be the first step in regulatory reform?
“Yes. Data is what reveals what people are thinking. However, because of the argument that we shouldn't know specific individuals' thoughts, data is being blocked. We need to clarify the definition of privacy, set clear standards, and release it. For instance, in healthcare, we have no access to patient records. While there are differences in levels, it’s much more freely utilized in foreign countries than here.
Regulations start from the logic of ‘Let’s just block it because we don’t know what to do.’ It’s about avoiding responsibility.
―How would you assess our workforce level?
“There aren’t many good talents in the country. There are many articles about semiconductor talent flowing overseas, and the same goes for software. The problem is how to stop this, but there are no realistic measures. In the U.S., AI talent earns more than $1 million, which is over 1.3 billion won; how can we handle that?
Currently, some corporations are establishing R&D centers in the U.S. to offer competitive salaries there. However, that technology isn't ours. While our corporations may utilize it, the talent is all being cultivated abroad.
―How can we bridge the talent gap?
“Among the written questions I received beforehand, there was an interesting analogy: ‘Moving from president of Samsung Electronics to Director General of R&D is like going from the coach of Real Madrid to the national team coach.’ How should we develop talent?” The first person that came to mind was Guus Hiddink.
In 2002, although each of our national team members did not reach world-class levels, Hiddink improved their fitness, built teamwork, and transitioned them into multifunctional players. We have to do the same. We have many research institutions and personnel. We need to enhance the physical capabilities of that workforce and unify them.
―Are you suggesting that we can solve the problem of insufficient talent through teamwork?
“Yes. If we proceed with R&D without strategy, both Research Institute A and Research Institute B will end up working on the same project. The same applies to corporations. After my appointment, when I visited a research institute, they were demonstrating something that Samsung had done 3 to 4 years ago. I couldn’t help but ask, ‘Why are they doing this?’ Someone needs to organize this. We need to clearly divide the areas where each can excel and manage the division of labor well. We need to work closely with corporations.
In the past, I believed that the Korean national soccer team was ‘physically strong and mentally tough,’ but Hiddink diagnosed it as ‘physically weak.’ If we reflect on what we think we do well, it does not translate into actual results. If we truly performed well, the outcomes should have manifested as products or services, or at least as papers, yet there is none.
―Collaboration between corporations and the government is important.
“Both corporations and the government need to align in their views on R&D. Both academic and industrial achievements are important. We need to change the practice of dividing R&D projects finely just to distribute them. The structure should shift from one person managing multiple small projects to several researchers working on substantially larger projects, allowing for collective intelligence to flourish.
―What level do you believe Korea's AI technology is at?
“I believe we are at a very nascent level. To understand why this is the case, it relates back to what I mentioned earlier about ‘years without innovation.’ There are discussions now about aiming to be one of the ‘three AI powerhouses,’ but if you exclude the frontrunner U.S. and the second-place China, the rest are all at roughly similar levels. In this context, being third has no significance. A fourth or fifth place is the same.
―What do you think about the national-led direction of AI development?
“We have a Chief AI Officer now, and I expect related policies will be forthcoming. The currently discussed sovereign AI (LLM-based AI developed without relying on foreign technology) could have significance as a public good, but it is unlikely to be a strategy that leads the entire industry. What we should prioritize is the development of industry-specific AI, that is, domain-specific generative AI. We need to create practical AI suited to each industry and establish a concrete roadmap.”
―Many point out that the pace is slow; what do you think is the reason?
“Regarding speed, I cannot overlook the national budget process. Currently, if a proposal comes in to invest in a particular industry, it takes two and a half years for it to be reflected in the budget. This year's budget is determined based on the previous year, and last year's budget was shaped based on demand identified two years ago. If discussion continues this way, the industry will have changed by the time the budget catches up.
―How should we transition?
“Look at corporations. Corporations move by first executing projects and then being held accountable later. However, the government starts with deliberation and makes decisions much later. I believe we need to shift to a ‘perform first, be accountable later’ system. The speed of decision-making and execution is the competitive edge in this era. The government's actions need to align with this.”
―What jobs will you prioritize?
“First, I plan to visit all domestic research institutes. I aim to understand the strengths of each institute and create guidelines for the division of R&D labor. I will also meet with many corporations to identify their needs and match them with research institutions. I see myself as a ‘mediator.’ Connecting both sides for effective collaboration is what matters most.”
―What do you envision for our status in five years?
“If we do nothing now, we could truly be miserable. My previous report projected a growth rate of 1.5% by 2034, yet there are already forecasts indicating it may drop below 1% this year. I celebrated too early; when I opened my eyes, I realized we’re in a situation where even ‘fast follow’ has become difficult.
Still, there is strength in our ability to unite when we feel a sense of crisis. I hope that time comes swiftly. In terms of cultivating talent, smart individuals need to return to engineering. For this to happen, an ecosystem that produces ‘sudden wealth’ must be created. We need to see more cases of becoming wealthy through startups.