Teaching software requirements engineering with event storming and generative AI: experience report

Abstract: This article presents an experience report with an exploratory approach, combining results measurement methods applied in the educational context of Software Engineering. The study aims to investigate the responsible use of Large Language Models (LLMs) in generating software requirements based on artifacts produced by Event Storming workshops. It aims to evaluate whether combining methodologies with AI-based tools can improve student autonomy and conceptual understanding. The research was conducted with final-year undergraduate students in a Systems Analysis and Development program. The activity was structured into three phases: training and project planning using Lean Inception, requirement generation using LLMs based on Event Storming outputs, and individual evaluation. We collected data through a Likert-scale questionnaire. The findings indicate that even students with no experience in Event Storming were able to use the technique to generate inputs for LLMs. The results showed a high utilization of the requirements generated by AI in conjunction with Event Storming. The combined use of Event Storming and LLMs has proven promising for teaching requirements engineering, fostering critical thinking, and promoting student autonomy. We observed in the study that students who had greater ease in meeting the requirements were more critical of the quality.