South Korea’s LG Group will introduce its first industry-use autonomous artificial intelligence (AI) agent, incorporating its large-scale AI model series EXAONE into the production lines of key affiliates in the second half of 2024.

EXAONE will be implemented across various facilities where electronic, chemical, bio, and communication products bearing the LG mark are produced.

From predictive and detection systems for equipment shutdown based on EXAONE to models detecting anomalies as the components are assembled, LG Group plans to transform its overall production lines into advanced AI smart factories, optimizing equipment processes in real-time, lowering costs, and increasing production capacity.

Pilot tests are currently underway within LG Electronics Inc., LG Household & Health Care Co., LG Chem Ltd., LG Energy Solution Ltd., LG Innotek Co., and other companies to build an advanced industrial autonomous AI agent at the LG Group level.

“If conventional production lines focused on smart factory levels within automation, EXAONE will take on the role of the agent’s brain in industrial settings,” LG AI Research vice president and research group leader Lim Woo-hyung said during a recent interview with Maeil Business Newspaper.

“We are continuously enhancing various AI models that are based on EXAONE, which are currently being evaluated on LG’s major production lines.”

“With this in mind, we plan to internally share optimized solutions for each affiliate’s production line in the second half of 2024,” Lim added, with this direction highly likely to be applied externally as well.

LG Group unveiled EXAONE 2.0 in July 2023, with the model the foundation for three specialized services ? EXAONE Universe (expert conversation AI platform), EXAONE Discovery (AI platform for new materials, substances, and drug development), and EXAONE Atelier (language and image extraction multimodal AI platform).

The AI models under research and development and slated for commercialization at LG AI Research are mainly supply chain management (SCM) optimization solutions for demand prediction, production process scheduling automation models, and anomaly detection solutions such as shutdown.

The pilot application of the process scheduling model for production optimization has been observed in LG Chem’s pharmaceutical manufacturing division.

“For example, we conducted pilot tests at an LG Innotek factory to automatically identify abnormal points in the production process and detect quality issues,” Lim said. “We are visiting numerous affiliate factories to create more efficient AI models for production processes.”

Regarding the anomaly detection model, Lim said that “During pilot tests at a chemical plant in the past, we felt the need for a solution that is in demand on-site. If a significantly large facility breaks down, it must be shut down for maintenance. But if a system can predict and schedule maintenance preemptively, it will significantly reduce the downtime.”

The focus of industrial AI solution development is shifting towards autonomous AI agents, with Lim noting that as production line scheduling is optimized, AI is already making decisions based on conditions and providing instructions to humans.

“As relevant technology advances, the role of autonomous AI agents will also grow,” he noted.

By Ko Min-suh, Seong Seung-hoon, and Lee Eun-joo

[ⓒ Pulse by Maeil Business Newspaper & mk.co.kr, All rights reserved]