A survey of 451 students at seven Thai Rajabhat universities extended UTAUT by positioning metacognitive awareness and self-regulated learning between technology-acceptance determinants and GenAI-use intention, with intention predicting self-reported use of GenAI as a cognitive partner. Facilitating conditions had the strongest associations with metacognition and self-regulation but no significant direct association with intention; the significant indirect path ran through metacognition. Social influence was the strongest direct predictor of intention, followed by metacognition, while effort expectancy and self-regulated learning did not predict intention in this model.
Key findings
- Facilitating conditions were associated with metacognition at β=.595 and self-regulated learning at β=.514, but not directly with intention, β=.061, p=.430.
- The significant indirect association between facilitating conditions and intention ran through metacognition, β=.186, consistent with indirect-only mediation in the specified model.
- Social influence was the strongest direct predictor of intention, β=.357, followed by metacognition, β=.312; effort expectancy and self-regulated learning did not predict intention. The model explained 62.1%–70.9% of endogenous variance.
Why this matters globally
The study moves higher-education GenAI research beyond whether students adopt a tool toward how institutional support and learner regulation relate to use. In resource-diverse university systems, guidance, support, and social norms may matter more than ease of use alone. Policy and curriculum interventions still require experiments or longitudinal evaluation, including learning quality, accuracy, academic integrity, and equity rather than intention alone.
Thai researcher contribution
Natthawan Phoonson, Thada Jantakoon, and Rukthin Laoha, working in the Rajabhat Maha Sarakham University context, tested the model with students from seven Rajabhat universities. This provides evidence from Thailand's regional higher-education system rather than only a large Bangkok university. The Thai contribution is the integration of UTAUT with metacognition and self-regulation in self-reported GenAI use, but it should not be generalized to all Thai students without stronger sampling detail.
Limitations to consider
All measures were collected at one time from the same respondents, creating common-method risk and preventing causal or temporal claims even though mediation was specified. GenAI use was self-reported rather than captured in logs or assessed for output quality. Reliability above .95 for six constructs indicates item redundancy, and the authors' full-collinearity diagnostics suggested possible common-method variance. Seven Rajabhat universities do not represent all Thai higher education, and no learning outcomes or skill assessments were tested.