Factors influencing university students' intention to use and reliance on generative artificial intelligence

Synthesis: Trang H. Nguyen, Long T. Truong, & Nhu H.T. Nguyen (2026) investigated factors influencing university students' intention to use and reliance on generative AI among Engineering and Computer Science/IT (CS/IT) students. Drawing on an extended Technology Acceptance Model (TAM), the study integrates the construct of critical use and conceptualizes reliance across four functional domains relevant to engineering and CS/IT education.

Key Findings

• Method: Anonymous survey at an Australian university (n=126), analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM).
• Attitudes and critical use drive intention: Attitudes toward GenAI and critical use directly and positively influence intention to use GenAI, while perceived ease of use and perceived usefulness have positive indirect effects. Notably, ease of use did not directly shape attitudes, and usefulness did not directly predict intention.
• Intention predicts reliance across four domains: Understanding, assessment, programming, and engineering projects. Students demonstrated moderate reliance overall, with greatest use for understanding-related tasks (clarifying concepts, worked examples) and limited utilization for full assessment writing.
• Appropriate, not excessive, reliance: Consistent with Schemmer et al.'s framework, students showed moderate, appropriate reliance — recognizing GenAI's limits for precision-demanding engineering and CS tasks while drawing on it for debugging, formula generation, and design ideation.
• Critical use as a core AI Literacy dimension: Students who actively evaluate, question, and cross-check GenAI outputs are more motivated to integrate the tools, and this practice safeguards against over-reliance.

What this means for practice

• Instructors. Teach critical use directly — students who interrogate, validate, and cross-check GenAI outputs were more motivated to use the tools, and this practice is the mechanism that guards against over-reliance.
• Instructors. Set task-specific expectations instead of one blanket rule: students relied most on GenAI for understanding-related tasks and programming, and least for full assessment writing, so policies and scaffolds should vary by task type.
• Faculty developers. Adopt the paper's three-tiered AI Literacy approach for engineering and CS/IT programs — evidence-based training for all students, mentoring or small-group support for those needing more, and intensive one-to-one support for students showing persistent difficulty or over-dependence.
• Administrators. Supply approved platforms, monitored environments, and clear acceptable-use guidelines, and use self-assessments or discipline-specific AI literacy checks to identify students at risk of uncritical use early.

Limitations

• The sample is small and single-site: 126 students from one Australian university, just above the minimum of 124 cases the authors calculated for this PLS-SEM model (power = 0.8, significance level = 0.05).
• Reliance and intention are measured by self-report survey, which risks response bias; students may have underreported GenAI use they saw as academically inappropriate because of social desirability.
• The institutional and disciplinary specificity of the sample limits generalizability to other disciplines, institutions, or national contexts.
• Data are cross-sectional, so the study cannot show how critical use and reliance evolve as GenAI tools change; the authors call for interviews, focus groups, or classroom observation to triangulate the survey.

Citation

Nguyen, T. H., Truong, L. T., & Nguyen, N. H. T. (2026). Factors influencing university students' intention to use and reliance on generative artificial intelligence: An extended technology acceptance model with critical use. Computers and Education: Artificial Intelligence, 10, 100618.