Examples of Scenarios for the Application of Explainable Artificial Intelligence in the Domain of Information Systems
In an era of increasingly intensive implementation of large language models within information systems, the issue of explainable artificial intelligence (XAI) has become crucial for maintaining trust, transparency, and ethical accountability. This paper examines methodological approaches to XAI in the context of both traditional AI and LLM-based models, focusing on their applicability, limitations, and role in meeting regulatory and ethical requirements. Through theoretical analysis, comparative evaluation of post-hoc and intrinsic methods, and practical insights involving models such as DistilBERT and GPT-5, the study illustrates how various XAI techniques can contribute to a deeper understanding of the behavior of complex AI systems. Particular emphasis is placed on five scenario-based examples of the development and testing of solutions in concrete organizational contexts. The application of XAI in information systems is further analyzed through the lens of credibility, stability, and usefulness criteria for explanations, as well as the challenges of integrating explainability into business information systems. The paper also addresses relevant regulatory frameworks and ethical implications that drive the advancement of XAI within the domain of information systems.