Enhancing User Story Generation in Agile Software Development Through Open AI and Prompt Engineering

This innovative practice full paper explores the use of AI technologies in user story generation. With the emergence of agile software development, generating comprehensive user stories that capture all necessary functionalities and perspectives has become crucial for software development. Every computing program in the United States requires a semester-or year-long senior capstone project, which requires student teams to gather and document technical requirements. Effective user story generation is crucial for successfully implementing software projects. However, user stories written in natural language can be prone to inherent defects such as incompleteness and incorrectness, which may creep in during the downstream development activities like software designs, construction, and testing. One of the challenges faced by software engineering educators is to teach students how to elicit and document requirements, which serve as a blueprint for software development. Advanced AI technologies have increased the popularity of large language models (LLMs) trained on large multimodal datasets. Therefore, utilizing LLM-based techniques can assist educators in helping students discover aspects of user stories that may have been overlooked or missed during the manual analysis of requirements from various stakeholders. The main goal of this research study is to investigate the potential application of OpenAI techniques in software development courses at two academic institutions to enhance software design and development processes, aiming to improve innovation and efficiency in team project-based educational settings. The data used for the study constitute student teams generating user stories by traditional methods (control) vs. student teams using OpenAI agents (treatment) such as gpt-4-turbo for generating user stories. The overarching research questions include: RQ-l) What aspects of user stories generated using OpenAI prompt engineering differ significantly from those generated using the traditional method? RQ-2) Can the prompt engineering data provide insights into the efficacy of the questions/prompts that affect the quality and comprehensiveness of user stories created by software development teams? Industry experts evaluated the user stories created and analyzed how prompt engineering affects the overall effectiveness and innovation of user story creation, which provided guidelines for incorporating AI-driven approaches into software development practices. Overall, this research seeks to contribute to the growing body of knowledge on the application of AI in software engineering education, specifically in user story generation. Investigating the use of AI technologies in user story generation could further enhance the usability of prompt engineering in agile software development environments. We plan to expand the study to investigate the long-term effects of prompt engineering on all phases of software development.