LG AI Research has unveiled “expert AI” models specialized in manufacturing, finance, and science, along with their deployment cases in industrial settings.

LG AI Research held “LG AI Talk Concert 2026” on the 14th at the Convergence Hall of LG Science Park in Magok, Gangseo-gu, Seoul. In his keynote speech, Im Woo-hyung, co-head of LG AI Research, stated, “Building high-quality AI models is important, but the mission of LG AI Research is to solve difficult problems that have remained unresolved in industrial fields for a long time,” adding, “To solve problems in real-world industrial sites, AI must be able to understand and judge countless variables and even the 1% of exceptions, and only by proving tangible outcomes can AI demonstrate its value in the field.”

Accordingly, LG AI Research has focused on developing expert AI equipped with domain-specific knowledge for each industry. It presented cases of achieving concrete results with AI in three major sectors: ▲manufacturing ▲science ▲finance.

In manufacturing, the institute introduced “EXAONE Tabular,” which predicts process conditions and quality information, and “EXAONE Omni-Inspect,” which performs inspections without retraining even when processes or product appearances change. Lee Soon-young, head of the Data Intelligence Lab at LG AI Research, and Yoo Jung-seon, head of AX at LG Innotek, explained that using industry-specialized AI foundation models enables fast and accurate predictions with only a small amount of field data and allows flexible responses to changing process conditions.

In science, Han Se-hee, lab head at LG AI Research, together with Park Je-seob, head of the R&D AX Promotion Team at LG Chem, introduced “EXAONE Discovery,” which designs materials and predicts synthesis outcomes. This model aims to build an autonomous experimentation system that carries out actual experiments. Han, together with Lim Jong-gu, head of the Fundamental Technology Team at GS Caltex, also presented a case study on developing immersion cooling oil materials for next-generation AI data centers.

LG AI Research also collaborated with LG H&H to review 420,000 candidate substances and, within a single day, discovered “Rhamsydil,” a substance effective for hair loss. With D&D Pharmatech, it is developing a next-generation oral peptide drug. Its proprietary model “EXAONE 4.5” has been embedded in the Ministry of Food and Drug Safety’s AI system for new drug review.

The institute further introduced “EXAONE Business Intelligence (EXAONE BI),” a finance-specialized AI solution in which multiple AI agents collaborate to generate outputs ranging from data analysis to reasoning, prediction, and explanation. Since its official launch early this year, EXAONE BI has been expanding its business with the London Stock Exchange Group (LSEG) for global markets and with Koscom for the domestic market. In particular, it incorporates “EXAONE Forecast,” an AI foundation model that understands information as it changes over time.

Since its establishment in December 2020, LG AI Research has addressed around 100 industrial grand challenges over the past six years, including predicting battery life and capacity, identifying defective products, optimizing production and materials management planning, and discovering drug candidates. During the same period, it presented 368 papers at domestic and international conferences and filed 1,080 patent applications.

The current status of developing a “Robot Foundation Model,” which serves as the brain of robots, was also introduced. LG AI Research stated that it is independently researching and developing safety-related algorithms capable of preventing accidents caused by robot operation in advance.

LG AI Research is also listed as a national flagship AI candidate in the “Independent AI Foundation Model” (“Dokpamo”) project promoted by the Ministry of Science and ICT. On the 18th of last month, it was selected, along with SK Telecom and Upstage, to advance to the next phase in the second-stage evaluation. Following a third evaluation, two final teams will be chosen early next year.

Popular News