In a world dominated by AI powerhouses like OpenAI, Meta, and Google, South Korea’s LG AI Research is stepping onto the global stage with a bold move—releasing EXAONE 3.0, the nation’s first open-source AI model. This groundbreaking initiative positions LG as a formidable competitor against industry giants, demonstrating its capability to hold its own in the race for AI innovation.
In an interview with The Chosunilbo on Nov. 27, LG AI Research President Bae Kyung-hoon said the decision to release EXAONE 3.0 as open-source was driven by confidence in its capabilities. “We believe our model can compete with global AI leaders in terms of performance,” Bae said. The model has excelled in areas such as coding and mathematics, even surpassing competitors in some cases.
Bae emphasized the importance of building an AI ecosystem through open collaboration. “If a single company monopolizes the field, reversing that hegemony becomes nearly impossible. That’s why sharing competitive AI models now is crucial,” he explained. He also revealed LG’s plans to release updated models every six months, mirroring the pace of major tech firms.
As LG AI Research approaches its fourth anniversary on Dec. 7, the institute reflects the vision of its leader, Bae Kyung-hoon. With a doctorate in computer vision and a global academic pedigree—including an MBA from Columbia Southern University and an executive program at Stanford—Bae has spearheaded its mission since its founding in 2020. His leadership was further recognized this year when he was inducted as a full member of the National Academy of Engineering of Korea.
LG AI Research President Bae Kyung-hoon poses for a photo after an interview with The Chosunilbo on Nov. 27, 2024./Kim Ji-hoInitially, LG AI Research aimed to develop massive AI models like Google’s and OpenAI’s. However, concerns over efficiency and practicality led the team to pivot toward lighter, specialized AI tailored for specific applications. This shift has already yielded significant results.
In March, LG AI Research partnered with the Jackson Laboratory, a U.S.-based genomics institution with ties to 20 Nobel laureates, to develop treatments for Alzheimer’s disease. The nonprofit initially reached out to LG for its specialized AI expertise.
“When we first received an email from Jackson Laboratory, we didn’t recognize the name and almost marked it as spam,” Bae said. “But their CEO, Lon Cardon, came to Korea personally. After considering partnerships with Google and Microsoft, they found our specialized AI better suited to their needs.”
This collaboration has since attracted interest from major pharmaceutical companies. LG AI Research is now working with two global firms to commercialize AI models for analyzing pathological images. “Our work with Jackson Laboratory has shifted perceptions, leading to a surge in inquiries from leading pharmaceutical companies,” Bae said.
While making global strides, Bae remains focused on integrating AI across LG’s operations. “Our goal is to solve real-world challenges and revolutionize productivity within LG,” he said.
One key achievement is the automation of a semiconductor production line at LG Innotek, which now operates without human intervention. “Even as a manufacturing-based company, LG has spent over five years organizing its data. Now, AI is helping turn that groundwork into measurable results,” Bae said.
Another initiative is enhancing AI proficiency among LG’s 80,000 office employees. This month, LG launched its enterprise AI agent, ChatEXAONE, to assist staff. For example, employees can use the system to analyze local consumer trends in Brazil and draft reports tailored to LG’s standards. “This will significantly boost productivity,” Bae said. “We aim to prove how a manufacturing company can lead AI-driven innovation.”
