How High-School Pressure and Autonomy Support Are Linked to Dual AI Learning Pathways: A Cross-Contextual SEM Analysis

Synthesis: Shen and Arunrugstichai propose a dual-pathway framework explaining why students' GenAI use ranges from constructive support to compulsive dependence. Drawing on Self-Determination Theory (SDT) and the Hook model, they examine how retrospective reports of high-school learning climate relate to university students' GenAI-use patterns and perceived AI-supported learning quality. Using cross-sectional survey data (overall N = 508; China-HS n = 305, non-China-HS n = 203; China-Uni n = 193, Thailand-Uni n = 315), they test the model with structural equation modeling and multi-group analysis.

Key Findings

• Dual pathways emerge linking learning climate to GenAI use: one toward constructive, autonomous use and another toward compulsive dependence.
• High-school pressure vs. autonomy support differentially predict which pathway students follow into university.
• Findings are consistent across cross-contextual groups (China vs. non-China high-school backgrounds; China vs. Thailand university contexts), supporting the framework's generalizability.
• The results link SDT-based motivational processes and behavioral mechanisms (Hook-model engagement) to perceived quality of AI-supported learning.

What this means for practice

• Instructors. Build autonomy-supportive conditions into GenAI tasks — meaningful choice, clear rationale, and low-pressure goals — because a pressure-laden high-school climate tracked with controlled motivation and compulsive AI use, while autonomy support tracked with adaptive use.
• Instructors. Support autonomous motivation and planned trigger readiness explicitly, since these are the motivational and behavioral links that connect autonomy support to adaptive GenAI use.
• Administrators. Treat incoming students' prior learning climate as intake information: it helps explain why GenAI use patterns diverge and where dependence risk is concentrated.
• Administrators. Monitor compulsive dependence as a distinct outcome from constructive use, with its own antecedents, rather than treating all GenAI use as one behavior.
• Researchers. Test the dual-pathway model — SDT-based motivational processes plus Hook-model engagement — as a joint explanation of productive and problematic AI use rather than studying either pathway alone.

Limitations

• The evidence is cross-sectional survey data from 508 university students in China- and Thailand-linked contexts (China-HS n = 305, non-China-HS n = 203; China-Uni n = 193, Thailand-Uni n = 315), so the SEM associations cannot establish that high-school climate causes later use pathways.
• High-school learning climate was measured through retrospective reports from university students rather than observed or concurrent measures.
• Outcomes such as perceived AI-supported learning quality, Well-Being, and academic performance are self-reported perceptions rather than objective records of use or attainment.
• Scalar-level measurement invariance supported cross-group comparison, but multi-group differences were concentrated in specific links and the context base is limited to the sampled China and Thailand groupings.

Citation

Shen, Q., & Arunrugstichai, W. (2026). How high-school pressure and autonomy support are linked to dual AI learning pathways: A cross-contextual SEM analysis. Computers and Education: Artificial Intelligence, 10, 100621.