Generative AI adoption in requirements-related activities in software product management : a qualitative study

Abstract

Generative AI (GenAI) tools have been studied extensively for their transformative impact on software engineering and product management in general; however, there is limited empirical evidence on how they are used in requirements-related work in software product management. This thesis adopts a qualitative approach to explore the application of GenAI in requirements activities, including requirements analysis, requirements verification and validation, and requirements specification (i.e., user story development and acceptance criteria definition), as well as the factors influencing its adoption. The research data for this study are drawn from eight interviews with industry practitioners to gather practical insights. The findings illustrate that GenAI is mainly used for language-intensive, documentation-oriented requirements tasks, where it assists in drafting, refining, and reviewing requirements artefacts and acts as a thinking partner during the requirements clarification and validation phases. Serving as an augmentation to existing practices rather than a human replacement. In contrast, strategic decisions and tasks involving personal accountability are excluded from AI use. Further investigation on GenAI adoption reveals its dependency on organisational governance, regulatory constraints, and individual preferences. The study concludes that GenAI serves as a cognitive partner and supportive technology in requirements-related work within SPM, with its integration remaining selective, bounded, and reliant on human oversight. This research presents empirical insights into the use of GenAI in requirements-related practice and the conditions influencing its adoption.