LLMS: ENHANCING MOBILE APP DEVELOPMENT EDUCATION IN A SOFTWARE STUDIO

INTRODUCTION: Large Language Models (LLMs) have become increasingly integrated into software development practices, particularly in programming activities. Their rapid adoption has created new opportunities and challenges for software engineering education, especially in environments where students are still developing fundamental programming skills. Although LLMs have demonstrated potential to support learning, there is still limited evidence on how they can be systematically integrated into Challenge-Based Learning (CBL) environments while promoting responsible and effective learning practices. AIMS: This research aims to implement the use of Generative Artificial Intelligence tools to enhance mobile application development education in a software studio. More specifically, the study proposes, implements, and iteratively evaluates the integration of LLMs into Challenge-Based Learning activities, investigating how students use these technologies and the perceived benefits and challenges associated with AI-assisted programming. MATERIALS AND METHODS: The study followed the Design Science Research Method (DSRM), in which the educational intervention, consisting of the integration of LLMs into CBL activities, was designed, implemented, refined, and evaluated through two iterative research cycles conducted at the Apple Developer Academy PUCPR, an educational program that adopts the Software Studio approach based on Challenge-Based Learning (CBL). At the end of each cycle, students produced reflective reports based on Schön’s Reflective Practice describing their experiences using LLMs during software development. A total of 100 reflective reports (50 from each cycle) were analyzed through an inductive qualitative approach using open and axial coding. The findings from the first cycle informed the refinement of the educational intervention implemented in the second cycle. RESULTS: The findings indicate that students initially used LLMs primarily to understand programming concepts and support learning, progressively incorporating them into broader software development activities as they gained programming experience. Students reported benefits such as personalized learning support, greater autonomy, increased motivation, and improved productivity. They also identified important challenges, including overreliance on AI-generated solutions, inaccurate responses, limited understanding of generated code, and concerns regarding excessive dependence on AI. Across the two DSRM cycles, students demonstrated a more selective, intentional, and critical use of LLMs. FINAL CONSIDERATIONS: The iterative implementation and evaluation of the proposed educational intervention provide empirical evidence that integrating LLMs into Challenge-Based Learning is a feasible strategy for enhancing mobile application development education while encouraging responsible, reflective, and critical use of Artificial Intelligence. The findings contribute to software engineering education by informing the design of educational interventions that integrate generative AI without compromising students’ autonomy, critical thinking, and learning processes.

KEYWORDS: Large Language Models; Programming Education; Mobile Application Development; Challenge-Based Learning; Software Engineering Education.