Motivational ecologies in AI ‐supported classrooms: Teachers and ChatGPT as dual agents

BACKGROUND: Generative artificial intelligence (GenAI) tools such as ChatGPT are increasingly integrated into classroom practice, yet their motivational-emotional implications remain insufficiently understood. Situated expectancy-value theory (SEVT) conceptualizes motivation and emotion as embedded in contextual affordances, raising the question of how AI-supported learning environments reorganize motivation. AIMS: This study examined motivational-emotional learner profiles in rubric-guided self-learning (SL) classrooms with and without ChatGPT access and tested whether perceived teacher motivational support and perceived ChatGPT assistance were associated with profile membership. SAMPLE(S): = 13.31, SD = 1.01) from non-AI-assisted SL classrooms (n = 709) and AI-assisted SL classrooms (n = 965). METHODS: Using a cross-sectional, person-oriented design, students reported interest, self-efficacy, effort, and negative emotions. Latent profile analyses were conducted separately by learning condition, followed by multinomial logistic regression. RESULTS: Four motivational profiles-Low Motivation, Medium Motivation, High Motivation, and one mixed profile -emerged in both contexts. Compared with non-AI-assisted classrooms, AI-assisted classrooms showed fewer students in Low Motivation (4% vs. 9%) and High Motivation (16% vs. 28%) profiles and a larger share clustered in intermediate configurations. Across contexts, higher perceived teacher support was associated with adaptive profile membership; perceived ChatGPT assistance provided additional differentiation in AI-assisted classrooms. CONCLUSIONS: ChatGPT access was associated with a redistribution rather than an overall increase in motivation, characterized by attenuated motivational extremes. The findings point to a hybrid motivational ecology in which teacher support remains central, while AI functions as a context-dependent, emotionally stabilizing affordance.