University students' acceptance of virtual teaching assistants in the era of AI: An extended UTAUT2 model

This study examines factors associated with university students' behavioral intention (BI) and self-reported use behavior (UB) regarding virtual teaching assistants (VTAs) using an extended UTAUT2 model. Data were collected through a cross-sectional online survey of 468 students with prior VTA experience at higher education institutions in Vietnam. The model incorporated personal innovativeness (PI), perceived VTAs value (PV), and perceived anthropomorphism (PA) alongside the original UTAUT2 constructs. Partial least squares structural equation modeling was used to test the hypothesized relationships, and SPSS 26.0 was used for descriptive analysis. All hypothesized associations were statistically significant. Performance expectancy was the strongest predictor of BI (β = 0.242; f² = 0.163). Perceived anthropomorphism had the largest standardized coefficient for UB (β = 0.314), whereas habit had the largest f² contribution to UB (f² = 0.168). The model explained 68.5% of the variance in BI and 60.9% of the variance in UB. The findings extend UTAUT2 to the VTA context and provide evidence from Vietnamese higher education. Because the study used cross-sectional and self-reported data, the findings should be interpreted as associations rather than causal relationships.