AI-Generated User Stories Supporting Human-Centred Development: An Investigation on Quality

User stories are an effective tool to document requirements in agile development, typically used to communicate user needs and technical details to developers in a non-technical language. As a result, poorly articulated stories often lead to misinterpretations, negatively affecting the development process as well as the quality of its outcome. Furthermore, the creation of user stories is often time-consuming, burdening project teams, where time is already scarce and tight schedules need to be met. This study sets out to explore the effectiveness of various Artificial Intelligence (AI) tools in generating high-quality user stories, based on customer or user interview transcript as input to the tool. Our approach is to use selected AI tools to generate multiple user stories. To ensure consistency and comparability, the same input prompt is provided to all AI tools. The Quality User Stories framework is then used to evaluate the quality of the generated stories. The application of these criteria enables a comprehensive evaluation of the syntactic, semantic and pragmatic properties of the stories. By comparing the results for the different tools, we draw patterns relating to each of the AI tools as well as a comparison of their performance, in terms of accuracy in extracted insights from the data. The significance of this research lies in exploring the extent to which AI can support the requirement specification process within the HCD lifecycle. AI tools could be embedded into the workflow, assisting practitioners in the analysis of transcripts to extract insights and create user stories. By having an assistive role in agile development, AI has the potential to save time and cost, enhancing overall efficiency and introducing more automation.