Impact of Generative Artificial Intelligence on Knowledge Management in Software Engineering: A Systematic Mapping Study

Abstract

Context: In recent years, Generative Artificial Intelligence (GenAI) has emerged as a transformative technology with high potential across various organizational contexts. Leveraging techniques such as Natural Language Processing (NLP) and Large Language Models (LLMs), tools like ChatGPT, GitHub Copilot, and DALL-E have begun to facilitate information creation and retrieval, automate tasks, and optimize decision-making processes. In the field of Knowledge Management (KM), GenAI has served as an enabler for organizing, disseminating, and reusing knowledge, fostering more collaborative and responsive environments. Specifically in Software Engineering, GenAI demonstrates potential to enhance productivity. The integration of GenAI with KM practices can play a key role in improving software quality, by enabling better reuse of knowledge, supporting consistent practices, and promoting informed decision-making. Understanding how GenAI can support KM in this domain is therefore essential to guide its strategic and effective integration. Objective: This study aims to investigate the impact of GenAI on KM processes within organizations operating in the Software Engineering domain. Method: To achieve this, we conducted a systematic mapping study to identify current practices, perceptions, and challenges related to the use of GenAI in KM within Software Engineering. Results: The study analyzed 97 primary studies published between 2021 and 2025. Main results indicate that GenAI focusing on code generation and software construction activities. KM practices most commonly supported include knowledge application and creation, with externalization emerging as the dominant SECI knowledge conversion mode. Conclusion: GenAI contributes significantly to KM in Software Engineering by supporting operational tasks and the formalization of knowledge. However, limited attention to knowledge dissemination and deeper learning processes reveals promising directions for future research.

Ähnliche Arbeiten