Integrating Generative AI into Knowledge Management: The Case of Software Development

The recent significant developments in Generative Artificial Intelligence (GAI) have contributed to renewing the interest of knowledge management (KM) scholars in Artificial Intelligence. In fact, the latest GAI tools, like ChatGPT or CoPilot, can process extremely large and varied sets of unstructured data and perform different tasks, such as classifying data, reading and writing texts, composing music, and developing software code. These tools already play a role in knowledge management by automating the generation of content and solutions. By harnessing GAI, knowledge management processes might become more efficient, scalable, and adaptive to the needs of organizations and users. In spite of this, studies on the use of GAI as a KM tool are still scarce, and there is a need to collect more empirical evidence about this phenomenon. This study intends to improve our understanding of GAI's use in KM by investigating how software developers use such tools in their knowledge generation process, i.e., to create new software code. The case of software developers was chosen because they are currently among the main adopters of this technology, and the first studies about the use of GAI in software development are already available. Furthermore, while software developers used to reach out to fellow developers on forums like StackOverflow, they now rely more on GAI for help with coding. The study, which, given the novelty of the topic, followed an exploratory approach, is based on interviews with 11 developers coming from8 different countries and having different backgrounds and experiences with GAI. We found that the interviewees liked the benefits of solving simpler programming tasks efficiently and rapidly. However, GAI requires expertise to review and modify the code and write the right prompts, as well as to evaluate the reliability of the output provided. The study also revealed that knowledge exchange with fellow programmers is partly, but not fully, exchanged with GAI, as it is more efficient. Furthermore, interviewees think that AI could become in the future a new member of the development team. Nevertheless, they also underline that continuous learning, adaptation, and ethical consideration are needed to fully realize the benefits of these powerful tools in software development.