Intelligent Management and Artificial Intelligence: Trends, Challenges, and Opportunities, Vol.2

Purpose: The aim of this article is to develop and verify an expert recommendation system based on ChatGPT (Generative Pre-Trained Transformer), which helps in selecting a heat source. Need for the study: With the development of large language models, the capabilities of artificial intelligence to process, interpret and infer from data are growing. These features are used in complex decision-making problems. One of these problems is the selection of a heating system. Methodology: The construction of the system included collecting data and preparing a domain knowledge base, followed by testing and fine-tuning. The verification of the developed system was carried out using case studies, including confirmation of the heat source selected by human experts or selection of a heating system. In addition, the performance and recommendations of the expert system were compared with the performance and recommendations of the original ChatGPT 4o. Findings: The research results indicate that the expert system, unlike the original ChatGPT, provided more precise calculations and more detailed data. Moreover, in the case of incomplete data, the system asked for details, reducing the risk of incorrect recommendations. Practical Implications: This type of ChatGPT-based expert recommendation system can effectively replace domain experts and positively influence the decisions made by individual stakeholders, businesses, and public institutions.