AI Must Go Beyond Models to Industry... Technology, Market, and Regulation Must Change Together

Private Sector Voices Emerge at Ministry of Science and ICT Roundtable
"Research Labs Unable to Buy GPUs, Medical Data Unusable... Let's Break Down the Barriers First"

Yonhap News Agency

"A foundation model is just an 'engine.' To become a true AI powerhouse, we must build the 'car' that the public can experience and that industry can use."(Kim Donghwan, CEO of FortyTwoMaru)

At the 'AI Industry Roundtable' hosted by the Ministry of Science and ICT and held at the Korea Artificial Intelligence and Software Industry Association (KOSA) in Songpa-gu, Seoul, on July 8, industry representatives voiced their opinions on a wide range of issues, including technology, application, regulation, and infrastructure. The meeting, presided over by Second Vice Minister Ryu Jemyoung, was attended by leading domestic AI companies such as LG AI Research, Naver Cloud, SK Telecom, KT, Upstage, FortyTwoMaru, and FuriosaAI.

The participants discussed a variety of topics centered on 'sovereign AI,' including continuous investment in foundation models for technological self-reliance, expansion of the AX (AI Transformation) ecosystem for the practical use of AI models, improvements to regulations, data, and power supply issues, as well as strategies for SME participation and global expansion.

Lim Woohyeong, Executive Director at LG AI Research, stated, "We believe it is important to build AI models that we can fully control, as this contributes to the domestic ecosystem. That is why we have been developing various technologies from scratch ourselves." He emphasized, "Now, what matters more than technology is service. We must quickly transition to an AI ecosystem that the public can experience and companies can actually utilize."

Lim also noted, "It is becoming increasingly important for not only large corporations but also the entire ecosystem of SMEs and startups to work together. Please help foster an environment where experimentation and failure can happen freely."

Bae Soonmin, Head of KT AI Future Lab, highlighted the importance of cloud adoption. He said, "Overseas, all corporate data is already on the cloud, so AI companies can apply it immediately. In Korea, however, we have to start with data refinement ourselves. Although we are late, this also means there is a significant opportunity."

Kim Donghwan, CEO of FortyTwoMaru, commented, "A foundation model is just the 'engine' of AI. With that engine, we need to build both passenger cars and buses. An AX ecosystem that can convert AI models into a variety of services must be established at the same time."

Yoo Hanjoo, Head of Naver Cloud Digital Healthcare Lab, specifically addressed the difficulties AI companies face in terms of regulation, policy, and infrastructure. He pointed out, "The Basic AI Act lacks the 'regulate after implementation' principle, so experimentation is blocked. Due to data center power regulations and uncertainty in licensing, it is even difficult to expand GPU capacity."

He added, "When trying to use medical data for AI training, in the United States it is allowed as long as the data is de-identified, but in Korea, if there is even a 0.1% chance of re-identification, it is blocked. Regulation that avoids risk without clear standards is holding back the industry."

Vice Minister Ryu Jemyoung stated, "AI determines national competitiveness and is becoming a key driver of 'real growth.' For this reason, the newly launched government is preparing to make it a core national agenda item. We will do our best to ensure that voices from the field are well reflected in national policy and that implementation proceeds smoothly."