LG showcases EXAONE results at world's top machine learning conference

LG Group unveiled industrial applications and research achievements of its in-house AI model EXAONE at ICML 2026, the world's foremost machine learning and AI conference. The company is accelerating its AI competitiveness by highlighting real-world use cases spanning new materials discovery, finance and data productivity.

LG said it participated in ICML 2026, which opened Monday at COEX in Seoul, presenting EXAONE's research and development results alongside industrial innovation cases. ICML, being held in Korea for the first time, is considered one of the three most prestigious academic conferences in machine learning and AI worldwide.

At the event, LG AI Research unveiled two new materials developed through EXAONE Discovery, its AI platform for new materials identification: Ramsidil, a hair-loss treatment compound being prepared for commercialization, and an immersion cooling fluid for next-generation AI data centers.

Ramsidil was co-developed with LG Household. AI identified it within a single day from among roughly 420,000 candidate compounds. The material has demonstrated hair-loss prevention effects without steroid-derived ingredients, with results presented at the World Trichology Society, and is currently being developed into a product.

The immersion cooling fluid was co-developed with GS Caltex, with plans to expand its applications going forward.

Earlier this year, LG AI Research completed patent registration covering the entire research and development process by which AI autonomously reads molecular structures from documents and designs target new materials, establishing a technological barrier to entry.

In the financial sector, LG AI Research introduced EXAONE Business Intelligence (EXAONE BI), a finance-specialized AI agent. The service analyzes approximately 8,000 stocks listed on Korean and US exchanges daily, providing predictive scores and research notes.

Earlier this year, LG AI Research launched a US stock prediction service with the London Stock Exchange Group (LSEG), and recently signed an agreement with Koscom to expand the project into the domestic market.

LG AI Research also unveiled EXAONE Data Foundry, an AI data factory platform that automatically generates high-quality data using AI and builds AI models tailored to specialized fields.

The platform has improved data productivity by more than 1,000 times compared with conventional methods and raised quality by an average of more than 20 percent. In a pilot project with the National Pension Service, the system automatically built more than 10,000 specialized data entries per day.

LG AI Research also presented research results, releasing a total of 14 papers at this year's ICML. Its generative AI technology for new materials ranked second globally on LeMat-GenBench, a leading international performance benchmark.

Since its founding in 2020, the institute has published a total of 363 papers at major global AI conferences including NeurIPS, ICLR and CVPR, and has filed 838 patents in Korea and abroad.

LG AI Research said it plans to continue identifying areas where AI can solve difficult challenges in industrial settings. "We will secure global AI leadership through a first-mover strategy aimed at solving hard industrial problems," said Lim Woo-hyung, co-director of LG AI Research.

During the conference, LG also held LG AI Day, inviting more than 60 master's and doctoral researchers from around the world. The institute's research staff shared updates on AI technology development and talent-nurturing strategies as part of efforts to attract top global research talent.

"AI leadership comes from the people who create the best technology," said Lee Hong-rak, co-director of LG AI Research. "We will build a system that provides researchers with the best possible environment for experimentation, data and mentoring, and leap forward as a world-class research hub."

Koo Kwang-mo, chief executive of LG Group, previously expressed his commitment to nurturing talent when addressing incoming students at the LG AI Graduate School — Korea's first officially accredited in-house corporate graduate school — saying he would be "the most steadfast supporter, so that you can freely unleash your imagination without being discouraged by failure, and so that the technology created here can meet the world."