Affective and Pedagogical Dynamics of ChatGPT Historical Simulations in Teacher Education: Insights from PANAS-Based Multivariate Modeling

This study examines the emotional and pedagogical impact of ChatGPT-based historical simulations in teacher training programs, exploring how AI-mediated learning environments shape students’ affective experiences and engagement. Drawing on the Positive and Negative Affect Schedule (PANAS), we measured a broad spectrum of emotional responses elicited during simulation activities across three subjects: Didactics of History, Didactics of Geography, and Innovation in Economics. The analytical framework combined descriptive statistics, non-parametric tests (Kruskal–Wallis and Mann–Whitney U), Spearman correlations, k-means clustering supported by Principal Component Analysis (PCA), and predictive modeling using Random Forests. The results show significant differences in emotional activation across gender, age, and subject, while highlighting the predominance of positive emotions—particularly Inspired and Attentive—as central predictors of emotional profiles. These findings suggest that affective responses, more than demographic variables, play a decisive role in shaping students’ interactions with AI-supported learning tasks. The study further contributes to emerging discussions in AI literacy and affective computing by evidencing how emotionally meaningful interactions with generative AI may enhance pedagogical engagement and support responsive instructional design