LG Corp. (003550.KS) unveiled a broad set of technologies built on its in-house EXAONE artificial intelligence models, including systems that inspect product quality on their own and a chemical laboratory where AI synthesizes and analyzes materials.
LG AI Research presented a range of "expert AI" models for manufacturing, science and finance at the AI Talk Concert 2026, held on the 14th at LG Science Park in Seoul's Gangseo district.
For manufacturing, the institute showed off EXAONE Omni Inspect, a model built for autonomous inspection. Earlier autonomous inspection models had to relearn data every time a production process or a product's exterior appearance changed. Omni Inspect performs high-accuracy quality checks without that retraining, LG AI Research said.
The new model will work alongside existing manufacturing models such as EXAONE Tabular, which interprets and forecasts process conditions and quality data. "An industry-specific AI foundation model can make fast, accurate predictions with only a small amount of on-site data, even in factories where conditions change often," said Lee Soon-young, head of the Data Intelligence Lab at LG AI Research, and Yoo Jung-sun, who leads AX operations at LG Innotek. "It can now respond flexibly to shifting process conditions."
In science, LG AI Research laid out a blueprint for an autonomous experiment system it is developing with LG Chem. The company plans to build a self-driving laboratory for chemical materials in which robots and AI, rather than human researchers, design compounds, predict the results of synthesis and run the experiments themselves.
LG plans to expand the business around EXAONE Business Intelligence, its finance-focused model, into North America, Europe and the Middle East. It will also widen the model's coverage beyond equity analysis to bonds and commodities. Since its commercial launch earlier this year, EXAONE BI has been growing through partnerships with the London Stock Exchange Group in global markets and with Koscom at home. The model, which handles analysis of listed companies, has been upgraded with EXAONE Forecast, an AI foundation model that interprets information as it changes over time.
LG AI Research also outlined progress on a robot foundation model, or RFM, intended to serve as the "brain" for robots that act in the physical world.
Since its launch in December 2020, LG AI Research has filed 1,080 patent applications and had 368 papers accepted, work it has used to solve more than 100 difficult problems on factory floors. Its results include predicting battery life and capacity, screening out defective products, optimizing production and materials planning, and identifying candidate compounds for new drugs.
"Over the past six years, LG has worked to make AI understand industrial problems, and now we are preparing for a stage where that AI judges and acts on its own," said Lim Woo-hyung, co-head of LG AI Research. "Building on the robot foundation model, we are moving beyond automating individual robots toward a future of autonomous factories, where an entire plant operates as a single intelligence."
LG says it intends to develop the EXAONE series as "expert AI for everyone" and deliver it to as many customers and users as possible. "The rules of technology and competition are changing, and customer expectations keep rising," LG Chairman Kwang-mo Koo said. "We have to move beyond the ways we have succeeded until now and leap into new innovation."
