Der Einsatz von Large Language Models in der Unternehmenswelt zur Kostenoptimierung im Supply Chain Management

Abstract

This thesis examines the integration of Large Language Models (LLMs) into business processes for cost optimization in Supply Chain Management (SCM). In view of the growing importance of artificial intelligence (AI) and the increasing cost pressure in SCM, the thesis aims to analyze the prerequisites and potential of LLMs to increase efficiency and reduce costs. The thesis is divided into several chapters. After the introduction, which defines the research questions and objectives, the second chapter describes the technology of LLMs and their technical and organizational requirements. The third chapter is dedicated to SCM, explains its objectives and costs and shows how LLMs can be used to reduce costs. The fourth chapter presents specific use cases and selection criteria for LLMs in SCM. The fifth chapter deals with technology acceptance by examining relevant models and influencing factors. Finally, practical recommendations for action are derived in the sixth chapter. The methodology includes a comprehensive literature review, case studies and empirical studies using questionnaires. The present work is dedicated to the investigation of the integration of LLMs into company processes for cost optimization in SCM. The central problem and the research questions derived from it are discussed and answered. The first research question deals with the technical and organizational requirements for the integration of LLMs and their use for cost optimization in SCM. Technically, LLMs require high-quality, well-prepared data, which is ensured by data cleansing, integration of data sources and feature engineering. Techniques such as prompt engineering, fine-tuning and retrieval augmented generation are necessary to adapt the models to company requirements. Organizationally, integration requires effective change management, training and continuous further training of employees as well as consideration of legal aspects such as data protection. The second research question examined the criteria for suitable use cases of LLMs in SCM. Suitable problems are characterized by data types, data availability, areas of application, process automation, business relevance and clear success criteria. The third research question investigated the factors that influence the acceptance of LLMs. Models such as the Technology Acceptance Model, the Unified Theory of Acceptance and Use of Technology and the AI device use acceptance model identify factors such as perceived usefulness, ease of use, social influence, expectation of performance and effort as well as hedonistic motivation and habit as crucial for technology acceptance. Derived factors for the survey show which factors are decisive in the corporate context. The recommendations for action in the work cover decisive points for the successful integration of LLMs for cost optimization in SCM, from data quality to user acceptance.

Date of Award	2024
Original language	German (Austria)
Supervisor	Veit Kohnhauser (Supervisor)