As the presidential election approaches, major candidates have presented key policy agendas related to artificial intelligence (AI), but they are facing criticism for lacking specificity. During the debate held on the 18th, Lee Jae-myung of the Democratic Party of Korea and Lee Jun-seok of the Reform Party engaged in a heated exchange over AI development measures.
Candidate Lee Jae-myung proposed '100 trillion won in AI investment' as his first pledge last month. He stated, "Through an AI investment of 100 trillion won, we will stand tall as one of the world's top three AI powers," and indicated plans to create a high-performance 'Korean version of ChatGPT' led by the government, making it available for free for all citizens to enhance productivity and achieve "an AI era where work-life balance is possible."
During the debate, candidate Lee Jae-myung responded to a question from candidate Lee Jun-seok, who asked, "To distribute accounts for a service like ChatGPT to all citizens, a budget of nearly 12 trillion won will be required. How will you approach this?" Lee Jae-myung replied, "By collaborating with private companies on research and development (R&D) to jointly develop an independent large language model (LLM). It may not cost 12 trillion won." He did not specify how much budget would be needed or how much of the funding for the sovereign AI development the private sector would bear.
He added, "It seems you are viewing this too pessimistically. Currently, users pay about $22 per month to use OpenAI's ChatGPT, and we will create an independent LLM based on domestic data, so that citizens can use ChatGPT for free, at least like using a calculator."
In response to concerns that the details of how the 100 trillion won pledge will be used are unclear, he stated, "It is not that we will raise all 100 trillion won from government finance; rather, we plan to attract private capital for R&D and sequentially secure investment funds each year."
Regarding the need for infrastructure to activate the AI industry, specifically the power needed to operate data centers, candidate Lee Jae-myung claimed, "We should transition to a society centered on renewable energy." Experts estimate that operating a high-performance AI data center requires power on the order of 1 gigawatt (GW). To achieve this, a power source that can provide a substantial amount of energy reliably and at a relatively low cost, like nuclear power, is essential, but Lee Jae-myung noted, "Nuclear power is fundamentally risky, posing sustainability issues."
Both the ruling and opposition parties have emphasized large-scale investment and nurturing the AI ecosystem in their main pledges, as they seek not to fall behind competitors like the United States, China, and France, which are ahead in the AI field. Experts have praised the candidates for recognizing the importance of the AI industry, but they pointed out that there are still no concrete implementation plans, making these pledges closer to declarations than actionable policies.
Kim Moon-soo, the candidate from the People Power Party, aimed for the country to emerge as one of the 'top three AI powers' by suggesting the nurturing of 200,000 young AI talents, establishing a 100 trillion won joint public-private fund to support AI unicorn companies, expanding the opening of public data, and creating roles for a Vice Minister of Science and Technology and an AI policy advisor. He stated that the power needed for AI operation would be supplied by nuclear energy. Candidate Lee Jun-seok revealed plans for attracting a data center to Pohang and relaxing regulations related to AI data training.
Experts evaluated that the candidates' AI pledges are promising in direction but lack an execution roadmap. The most critical elements needed to foster the AI industry are funding and excellent personnel, but up to now, there have been no details defined regarding funding sources, the level of private burden, plans for nurturing skilled personnel, power supply methods, or investment priorities.
First of all, there is no determination of how to use the 100 trillion won, and there are concerns that if the budget is not allocated efficiently, it could lead to waste of taxpayer money. Choi Byeong-ho, a professor at Korea University’s Artificial Intelligence Research Institute, remarked, "From the perspective of AI technology development, the more investment money, the better, but if substantial amounts are injected without a precise direction, the efficiency of policy implementation could decrease."
Considering that the government's budget this year is 677 trillion won, with an R&D budget of 29.6 trillion won, analysis suggests that even attracting private funds will not easily allow for securing 100 trillion won.
Experts have emphasized that measures must be established to prevent the outflow of AI talent. Given the current salary gap between high-level AI-related jobs in domestic and overseas corporations is over ten times, there is a pressing need for policies that can significantly improve the treatment of domestic AI talents.
Kim Myung-joo, the head of the AI Safety Research Institute, stated, "In our country, the outflow of AI talent is a bigger problem than nurturing them," and added, "It is difficult to completely prevent the outflow to big tech and other foreign companies that favor AI talent, and we also need to consider ways to attract competent talents from countries like India and Southeast Asia."
In terms of nurturing excellent personnel, some opinions suggest that expanding the relevant R&D budget would be more effective than establishing new AI departments. An industry insider noted, "Each school already has computer science departments that can develop AI talents, so it raises questions about the necessity of creating new AI departments," criticizing it as "a typical showpiece policy." Particularly in the current climate where medical school is preferred and STEM fields are avoided, there are concerns that even if new AI departments are established, it would be challenging to train the needed personnel in time.
The industry welcomes the large-scale investment pledges. However, they have assessed that rather than focusing on abstract concepts like 'AI foundational society,' it is urgent to implement regulatory relaxation for building the foundational large-scale power supply needed for AI deployment. An industry representative stated, "High-performance AI data centers are equipped with a large number of graphics processing units (GPUs), which consume tremendous amounts of power," adding, "Currently, our country lacks sufficient power to operate data centers."
The Korea Chamber of Commerce and Industry stated in its policy proposal presented earlier this month, "The next 3 to 4 years is a golden time for our country to rise to an AI G3 nation, so national capabilities should be focused on the AI field."