Development of a Decision-Making Method for Using Large Language Models in Production-Related Processes

Abstract Large language models (LLMs) have demonstrated significant potential across various industries. Despite their success in areas such as customer service, sales and marketing, their application in manufacturing and production environments often lacks clear, practical guidance for identifying suitable use cases and making effective implementation decisions. This study addresses this gap by developing a structured method to identify production-related processes suitable for LLMs and to guide implementation decisions. The research begins by outlining the fundamentals of LLMs. Practical examples from literature and industry are presented to illustrate the current state of LLM use in production-related contexts. Based on these insights, a two-part decision-making method is introduced. The first part focuses on identifying processes that could benefit from LLM integration. Criteria are developed through qualitative content analysis of the presented examples, and an evaluation process inspired by utility value analysis is applied. The second part employs a structured assessment framework to guide implementation decisions, assessing factors such as the appropriate deployment type (e.g., on-premises or cloud-based), the optimal size of the LLM, and knowledge transfer requirements. The questions are based on the previously outlined fundamentals of LLMs. The practical application of the method is demonstrated through a detailed case study, showing its effectiveness in identifying suitable processes and supporting practical decision-making as well as the need for further refinement.