[Report] "Farewell to Wrist Pain from Scraping Slag" POSCO's 'Managers' Embrace AI [AI Era, Jobs Are Changing]

Full Implementation of AI-Based KR Automation System
Introduction of Quantitative Metrics such as 'Slag Removal Rate'
Enhancing Quality and Safety

Editor's NoteThe use of artificial intelligence (AI) in industrial sites, including manufacturing, is expected to be the biggest topic in the industry this year. AI, which has already been deployed on manufacturing floors, has proven its role as a driving force for productivity improvement. This year, its scope will expand in earnest. However, behind this AI-driven production innovation lies concern over job losses. This is because the concept of the so-called 'Dark Factory,' where production is carried out without human intervention, is gradually becoming a reality, making full automation seem achievable. However, The Asia Business Daily's investigation into job changes at manufacturing sites where AI has been applied reveals a more complex picture. Repetitive and hazardous tasks are now handled by machines, while humans are clearly taking on roles of judgment, management, and responsibility, resulting in a clear division of labor. Rather than replacing people, AI is beginning to redefine jobs. The AI era has brought not only productivity and technological competition but also the challenge of job transformation. The Asia Business Daily visited industrial sites in the new year to directly observe the changes in jobs brought about by AI.

POSCO Pohang Steelworks 3rd Steelmaking Plant 'Pre-treatment (KR) Control Room' in Pohang, Gyeongbuk. The characteristic roar and intense heat of a steel mill are nowhere to be found; instead, dozens of monitors, reminiscent of an IT company's control center, fill the silence. On the split screens, a massive skimmer is meticulously scraping slag-the residue floating atop molten steel. This is the very site where, in the past, skilled workers would visually check the furnace flames and operate a joystick by hand.

At the main console in the control room, artificial intelligence (AI) was analyzing on-site data in real time. The control screen dynamically displayed the converter number, the composition of the molten steel, the location and operational status of the pre-treatment equipment, and the movement trajectory of the skimmer.

Last month, at the POSCO Pohang Steelworks 3rd Steelmaking Plant 'Pre-treatment Control Room' in Pohang, Gyeongbuk, a worker was observing the slag removal operation performed by artificial intelligence (AI). POSCO

Freedom from Shoulder and Wrist Pain Thanks to AI

The pre-treatment process, which marks the starting point of steelmaking, involves removing impurities from molten steel coming out of the furnace and adjusting the sulfur (S) content, a key factor in quality. This is called the 'slag removal' operation. Although it may seem simple, adjusting the tilt of the ladle (the vessel for molten steel) to remove only the slag in front of high-temperature molten steel requires intense concentration.

Shin Seungmin, a supervisor in the steelmaking department, said that after the introduction of AI, "my shoulders feel much lighter." Previously, he had to operate the joystick more than 400 times for each task. On days with high production, his wrists and shoulders would ache.

However, since AI was introduced to all pre-treatment facilities at the 3rd Steelmaking Plant last May, Supervisor Shin has been able to focus on monitoring the overall quality of the process rather than manipulating the joystick. He said, "With the reduction in physical burden, I can now concentrate on new tasks, such as equipment management and coming up with ideas to improve the system and the overall work environment."

'From Instinct to Indicators'... World's First Quantitative 'Slag Removal Rate' Introduced

During the preliminary processing stage, artificial intelligence (AI) is directly judging and removing slag. POSCO

The core of this innovation lies in the combination and use of thermal imaging cameras and high-precision angle data. Kwon Ohyeong, assistant manager in the steelmaking department, explained, "As we automated the know-how of highly skilled workers, the camera recognized the distribution of slag, allowing us to use an objective indicator called the 'slag removal rate'." In other words, AI was applied to the traditionally instinct-driven slag removal process to create measurable indicators. The ratio of removed slag and the iron (Fe) loss rate are now collected as real-time data, making this the first case of introducing a quantitative metric to a traditionally skill-based operation. This has laid the foundation for precise quality management by adjusting the slag removal rate according to the steel grade. The time required for slag removal has also been reduced by about 3-5% compared to before.

AI has also solved the most technically challenging step: 'tilting the ladle.' To remove slag effectively, the ladle must be tilted at the right angle; if the angle is off, molten steel can overflow, leading to major accidents. To prevent this, POSCO established a high-resolution camera system and a 'redundant decision algorithm.' If any abnormal signs are detected, the AI automatically stops the operation-a double safety mechanism.

Technology That Protects, Not Replaces, People

Since last year, the adoption of AI in domestic manufacturing sites has begun in earnest. Along with the implementation of autonomous manufacturing systems, concerns have arisen that on-site personnel may lose their place. However, POSCO has demonstrated that coexistence is possible by combining the know-how of skilled workers with AI. The result is the realization that while AI is responsible for data collection and learning, it is the skilled workers on site who generate that data.

The roles of on-site employees have not disappeared but evolved. Instead of simple repetitive work, training has been provided to develop capabilities as 'automation system managers.' Workers participated directly from the development stage.

On-site, the transfer of systems incorporating the know-how of highly skilled workers is already actively taking place. Assistant Manager Kwon said, "We educated workers about the structure of the pre-treatment automation system and how to respond in emergencies, focusing on the role of automation system managers. From the moment development was completed, we invested a lot of time and effort in writing user manuals with developers and operators."

Yoon Jeonggyun, head of POSCO's steelmaking department, said, "In the early stages of system implementation, there were many shortcomings because we could not fully replicate the workers' delicate operational methods, but as field workers started to participate in regular meetings, we were able to develop an automation system with capabilities similar to those of the workers." He added, "While the scope of AI, such as robotics and pre-treatment automation, is expanding, efforts to ensure worker safety are also important."