POSCO E&C has launched a company-wide AI challenge aimed at discovering innovative applications of artificial intelligence (AI) to transform construction site operations. Notably, an AI model for automating work report creation has been highlighted for its potential to save hundreds of hours of labor annually.
On May 17, POSCO E&C announced the conclusion of its "Company-Wide AI Challenge," which began on March 24. The competition involved all employees, including field staff, who engaged in learning and applying AI technologies directly to their work.
The challenge featured four categories: video, reports, AI agents, and a golden bell quiz, with nearly half of the workforce, totaling 1,887 participants.
The standout achievement was the top entry in the AI agent category, the "Work Report Automation AI Agent." Previously, employees from partner companies communicated work details via social media, requiring site managers to manually compile these into daily reports.
This AI agent automates that process, reducing repetitive tasks that previously took over 90 minutes each day. POSCO E&C estimates that this translates to approximately 375 hours of labor saved per site manager annually.
In the report category finals, participants were tasked with using AI in real-time to produce executive-level reports without prior topic disclosure. POSCO E&C stated this approach validated practical AI application skills among employees.
A company representative noted, "The greatest achievement of this competition is that employees have begun to see AI not as a technology unrelated to them, but as a 'colleague that changes their work.' We have laid the groundwork for a smart work culture that reduces repetitive tasks and allows for a focus on more essential duties through AI-driven innovation."
However, challenges remain in training AI due to the prevalence of informal language, jargon, and typographical errors in the unstructured data typical of construction sites. To improve the quality of AI training data, efforts are needed to standardize scattered site data, along with the establishment of government guidelines.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.