Automated extraction of competencies for assembly operators from process and product information using a large language model

In manufacturing, despite widespread automation, human labor remains indispensable due to its inherent flexibility. Effective workforce utilization demands precise identification and assessment of competencies for manual tasks. Traditional competency identification, typically done by human experts, is labor-intensive and subjective, often involving repetitive efforts without the benefit of reusing analyses across different jobs and tasks. This research introduces a methodology using a large language model to automatically extract required competencies from process and product data. Real-world experiments demonstrate a significant reduction in manual labor, presenting an open framework for continuous improvement in competency identification and management in manufacturing operations.