STRUKTURIRANA UPOTREBA GENERATIVNE VJEŠTAČKE INTELIGENCIJE U SOFTVERSKOM INŽENJERSTVU: ANKETNA ANALIZA I FUZZY-GENETSKI MODEL ZA PEDAGOŠKU INTERVENCIJU
STRUCTURED USE OF GENERATIVE ARTIFICIAL INTELLIGENCE IN SOFTWARE ENGINEERING: A SURVEY ANALYSIS AND A FUZZY–GENETIC MODEL FOR PEDAGOGICAL INTERVENTION

Summary/Abstract: The rapid and widespread adoption of generative artificial intelligence in software engineering education necessitates a balanced instructional approach—one that leverages the benefits of these tools while preventing excessive student dependence on automated outputs. This study integrates a student survey (N = 79) with a hybrid fuzzy–genetic (GA) decision model that quantifies the need for pedagogical intervention as a function of usage frequency and perceived risk. Nonparametric analysis reveals statistically significant differences related to career orientation (H = 14.192; p = 0.028) and usage-frequency patterns, with higher-intensity users employing generative AI across a broader set of tasks, including learning programming languages, generating solutions, and developing prototypes. The fuzzy inference system—constructed using nine rules and trapezoidal/triangular membership functions—produces a defuzzified output on a 0–10 scale (“low” to “urgent”). A genetic algorithm further optimizes membership boundaries and decision thresholds. The optimized model yields a low mean prediction error and reliably identifies “high exposure + (moderate or high) risk” scenarios as priority cases for pedagogical intervention. This supports an AI-enhanced projectbased learning (PBL) paradigm in which large language models automate routine tasks, enabling students to focus on validation, evaluation, and solution design. The findings provide a robust empirical and computational foundation for the systematic, controlled integration of artificial intelligence into the curriculum. They also underscore the importance of long-term, cross-course and cross-institutional monitoring of learning-related impacts.

• Details
• Contents