Deputy Prime Minister Bae Kyung-hoon: “Gyeongnam, a Core Hub of the ‘5 Poles and 3 Specialized Zones’ Strategy with Physical AI”

According to Financial News, Deputy Prime Minister and Minister of Science and ICT Bae Kyung-hoon stated, "We will combine physical artificial intelligence (AI) with Gyeongnam’s industrial competitiveness and firmly establish the region as a key hub in the ‘5 Poles and 3 Specialized Zones’ strategy for regional manufacturing."
On the 6th, Bae visited Shinsung Delta Tech in Changwon, South Gyeongsang Province (Gyeongnam), to review the results of the preliminary verification project on physical AI. At the site, he said, "Gyeongnam offers an optimal environment, with concentrated capabilities in precision manufacturing for machinery, parts, and equipment."
He went on to say, "Physical AI is a core task that will determine the future of Korea’s manufacturing competitiveness," adding, "At this stage, the most important thing is to systematically build data generated on actual manufacturing sites."
Bae continued, "Once a physical AI foundation model is developed based on this data, we will spread it in an open source format and build an ecosystem where many companies can easily make use of it." He emphasized, "The government will not stop at short-term achievements. With a sense of responsibility for Korea’s future, we will continue to push forward physical AI policies."
The visit was arranged to verify the applicability of physical AI-based precision control technologies in the field and to hear from companies and researchers about how to link these efforts with the upcoming large-scale Gyeongnam AI Transformation (AX) research and development (R&D) project.
The Gyeongnam AX project, which the Ministry of Science and ICT (MSIT) is launching this year, focuses on directly embedding on-site physical characteristics and skilled workers’ know-how into AI models. The project centers on developing large action model (LAM)-based physics-informed neural network (PINN) technologies that allow AI to directly control robots and equipment. This represents a shift from conventional "analysis- and judgment-focused AI" toward AI that "knows the shop floor best and actually runs the production process."
MSIT and the National IT Industry Promotion Agency (NIPA) are carrying out a preliminary verification project using the 2025 supplementary budget. They conducted on-site demonstrations at eight manufacturing companies in Gyeongnam.
These demonstrations produced visible results at major participating companies, including better prediction of process quality and improved production efficiency. At Shinsung Delta Tech’s plastic injection and assembly process, an AI training dataset was built by linking 49 types of injection molding process data with 62 types of action data, such as worker behavior, raw material conditions, and defect patterns. Using a digital twin model to predict and correct process quality in advance, the company confirmed the potential to reduce defect rates by about 15% and increase equipment utilization by about 20%.
Hwasung R&A improved its overall equipment effectiveness by more than 5% by predicting material deformation in rubber extrusion processes in advance. CTR predicted chattering, or machine vibration, occurring during aluminum machining, which reduced defect rates and shortened machining cycle time by more than 17%.
MSIT will fully launch the Gyeongnam AX project in the first half of this year. Running through 2030, the project will focus on realizing ultra-precise control physical AI by developing physical-intelligence action model technologies based on on-site manufacturing data.
At the on-site meeting, company representatives discussed the spread and application of physical AI-based precision control technologies, data management, and the modeling of skilled workers’ know-how. They also called for stronger policy linkages at the government level.
Gyeongnam is a region where manufacturing bases are concentrated around advanced industries. Because it offers favorable conditions for accumulating and testing field-centered data, MSIT plans to actively reflect the opinions raised in designing future regional AX initiatives and in shaping policy support so that demonstration results can lead to industrialization.
mkchang@fnnews.com Jang Min-kwon Reporter