“With LG Exaone, we will create an AI that leads rather than follows within two years.”

As leading domestic companies form a consortium to develop a global top-level AI foundation model, Lim Woo-hyung, Co-Director of LG AI Research Institute, stated this in a recent interview with this publication. Despite the overwhelming dominance of the US and China in the AI frontier model competition, he expressed confidence that Korea, ranked third globally, can also create a top-level AI model.


In fact, tangible results are emerging, such as the recent evaluation by Microsoft's 'AI Diffusion Report,' which compared representative AI models from various countries and ranked Korea's LG Exaone 4.0 model third after OpenAI's GPT-5 from the US and DeepSeek 3.1 from China. With GPT-5's performance rated at 100, DeepSeek 3.1 scored 84.1, and Exaone 4.0 scored 82.4. Additionally, the report assessed the technology gap with the US as 5.3 months for China and 5.9 months for Korea.

Lim expressed optimism, stating, “By the time the Ministry of Science and ICT's 'National AI' project concludes in 2027, most public institutions in Korea will be using LG Exaone.” Regarding the current situation where most domestic developers are already utilizing open-source AI models from the US and China, he pointed out, “When using Alibaba's Q-One model from China, we cannot know what training data it was trained on,” adding, “Even with additional tuning, the issue remains whether we can control and trust it.” He explained, “The strength of AI models developed in Korea is that we can predict how they operate because we control the variables from the start, and we can manage and improve them if issues arise.”

Having competitiveness in specialized domains such as manufacturing and bio is also considered an advantage. Lim noted, “The US lacks manufacturing data,” and emphasized, “When telling global clients to use 'Exaone,' we highlight that it performs better in manufacturing and bio domains compared to general US-China AI models.” With LG's unique data and AI technology that even big tech companies envy, the plan is to make a decisive move in specialized AI areas such as manufacturing, bio, materials, and finance.

It has already become a reality to produce over 5,000 analysis reports on US-listed companies daily based on Exaone. LG AI Research Institute has embarked on full-scale commercialization with an AI-based market prediction model developed in collaboration with the London Stock Exchange Group (LSEG). This model, known as the 'AI-based Equity Forecasting Score (AEFS),' analyzes 5,000 stocks listed on the New York Stock Exchange daily to provide investment insights.

LG AI Research Institute is also preparing to establish an 'automated laboratory' that synthesizes and analyzes new materials designed by AI automatically. In this laboratory, robotic arms will automatically synthesize new materials designed by AI and analyze experimental results in real-time to design re-experiments independently. The plan is to create a virtuous cycle that quickly develops optimal materials. Lim emphasized, “Now, the imagination of AI designing new materials and robotic arms synthesizing them is becoming a reality,” adding, “The innovation of LG AI Research Institute will be a significant turning point in changing the future of the industry.”

Popular News