[The National AI Team Trapped by the Independence Dilemma] "Reference Allowed"... Government Proposes a Compromise ①

"Independent K-Architecture Is Not Easy"
Realism Emerges
Government: "Open Models Are Allowed... Must Prove Unique Technological Capabilities"

Editor's NoteThe government’s “Independent AI Foundation Model (Dokpamo) Project,” often referred to as the national team selection for artificial intelligence (AI), is facing heated controversy over its “independence.” Critics argue that the “from scratch” requirement in the first evaluation round was not clearly defined, fueling calls for the government to clarify its position. The Asia Business Daily examines the industry’s perspectives on the independence debate from multiple angles and explores ways for the Dokpamo Project to become a stepping stone for the advancement of Korea’s AI ecosystem.

"It is permissible to use global open models. However, during the development process of an artificial intelligence (AI) model, including pre-training, we must clearly demonstrate our own differentiated development capabilities." (Senior official at the Ministry of Science and ICT)

Amidst a backlash in the AI industry over the independence criteria during the first selection phase of the government’s “Independent AI Foundation Model” project, the government has provided a clear interpretation of what constitutes independence. As criticism mounted-especially from developers and AI startups-over allegations that some elite team candidates were “copying” Chinese models, calls for clearer standards grew louder, prompting the government to issue a set of guidelines.

Even with the second round of selection approaching, debate continues over the independence criteria. Some argue that, since the government will invest significant funding and GPU resources in the final teams, a truly independent and unique architecture must be a prerequisite. However, others counter that, given the prevalence of open models based on proven large language models (LLMs) from the United States and China, it is not realistic for latecomers to design a completely original Korean architecture from scratch.

Complete Independence vs. Practical Use of Open Models: Government Offers Compromise Amid Controversy

During the first selection, all five finalists (LG AI Research, Upstage, SK Telecom, Naver Cloud, and NC AI) differed in degree but did not present entirely independent model designs. Even LG AI Research’s “Exaone,” which received the highest score, was not free from controversy, leading the developer community to call for more robust evaluation standards from the government. As the debate over independence persisted, the government took the unusual step of reopening applications beyond the three elite teams initially selected (LG AI Research, Upstage, and SK Telecom), with Motif Technologies and Trillion Labs entering the competition by emphasizing their unique architectures.

On January 27, a senior official from the Ministry of Science and ICT stressed, "Institutions applying for the program must prove their AI model development capabilities and experience-from model design, such as loss logs, to pre-training-by providing results from the latest global benchmarks and major leaderboards, such as those managed by Artificial Analysis, a leading international AI research institute."

A representative from the AI industry, speaking on condition of anonymity, commented, "It appears that all five companies in the first round started by referencing different Chinese architectures, and some models adopted those structures without much modification. While LG AI Research’s Exaone, which placed first, cannot be considered entirely original or independent, nor does it feature a completely new concept, it did attempt significant optimization and showed new combinations compared to competitors."

President Lee Jaemyung also appeared to acknowledge the independence debate in the AI industry. On January 25, he shared an analysis by Artificial Analysis on his X (formerly Twitter) account, which stated, "Several Korean AI research institutes have emerged at the cutting edge, thanks to Korea’s national AI initiative," and encouraged the industry by quoting, "If you can’t avoid it, enjoy it." Lee Jaehui, Professor of Computer Science at Seoul National University, remarked, "In the LLM industry, it is common for competitors to learn from and imitate each other’s work to continually improve performance. In an environment where the use of open models is standard, it is unreasonable to place excessive emphasis on independence."

A senior government official added, "The government fully recognizes that it is difficult to create a completely independent model. However, with foreign LLMs dominating the domestic market, building sovereign AI is extremely important. If fierce competition leads to improved capabilities and ultimately strengthens Korea’s AI ecosystem, it will be a meaningful achievement."