AI Ed Wiki logoAI Ed Wiki
Use with AI

📄 Research Article

Exploring AI perceptions in education: unveiling the role of student and teacher motivation and self-efficacy

● high · 2026-08-16 · student-experience, motivation, self-efficacy, k-12, ai-literacy, equity-in-ai-education

Synthesis: Baez, Buchner, Ullrich, and Schallert-Vallaster (2026) explore how primary and lower secondary school students perceive AI in learning, and how student and teacher Motivation and Self Efficacy relate to those perceptions. Using questionnaire data collected in Spring 2023 from 907 students (466 female, 441 male; 639 primary, 268 lower secondary; mean age 11.61 years, SD = 1.85) and their 53 class teachers (38 female, 15 male; mean age 39.32, mean 14.43 years of experience) clustered across 36 primary and 17 lower secondary classes in 12 primary and 3 lower secondary schools belonging to five school units in German-speaking Switzerland, and analyzed with a doubly multilevel structural equation model (ML-SEM, MLR estimator, lavaan in R), the authors found that students' intrinsic motivation to learn with digital media is significantly linked to their perception of AI at the individual level (β = 0.29, p < .001), with the model explaining 47% of the variance in AI perception. Students' self-efficacy is positively associated with their motivation (β = 0.43, p < .001), while girls exhibited lower self-efficacy for learning with digital media than boys (β = −0.17, p < .001). At the class level, teacher motivation to integrate digital media was positively associated with aggregated student motivation (β = 0.71, p < .01) and strongly linked to teacher self-efficacy (β = 0.86, p < .001), whereas teacher self-efficacy showed a negative association with class-level student motivation (β = −0.63, p < .05) once teacher motivation was controlled.

Key Findings

Motivation drives AI perception. Students' intrinsic motivation to learn with digital media was significantly linked to their perception of AI at the individual level (β = 0.29, p < .001), with the multilevel model explaining 47% of the variance in AI perception; the three-item AI-perception scale (α = .87) asked students whether a machine with AI could help them when stuck, would be practical, and whether they would like to use it.

Self-efficacy feeds motivation, with a gender gap. Students' self-efficacy to learn with digital media (3-item scale, α = .75) was positively associated with their motivation (β = 0.43, p < .001), but girls showed lower digital-media self-efficacy than boys (β = −0.17, p < .001) — a gender gap consistent with prior ICT self-efficacy meta-analyses and relevant to Equity In AI Education.

Age matters at the class level, not the individual level. Individual student age was not significant, but older classes showed lower aggregated motivation to learn with digital media (β = −0.91, p < .001), mirroring the well-documented decline in intrinsic motivation from 3rd to 8th grade.

Teacher motivation transfers to the class. Teacher motivation to integrate digital media (3-item scale, α = .84) positively predicted class-level student motivation (β = 0.71, p < .01) and was strongly linked to teacher self-efficacy (α = .87; β = 0.86, p < .001); notably, teacher self-efficacy was negatively associated with class-level student motivation (β = −0.63, p < .05) when teacher motivation was controlled — a counterintuitive result the authors attribute to highly self-efficacious teachers possibly implementing more challenging instruction.

Multilevel design. A doubly latent ML-SEM on a large Swiss sample (907 students, 53 teachers; 12 primary and 3 lower secondary schools in a school-development program focused on digital-media integration) provides robust evidence. AI perception showed no meaningful between-class variance (ICC below 0.1) and was therefore modeled only at the individual level, while student motivation was decomposed into within- and between-class components; the model fit well (CFI = 1.00, RMSEA = 0.00).

Implication. Fostering students' self-efficacy and motivation — especially among female students — plus teacher AI Teacher Education (e.g., Tell-Show-Enact-Do learning designs) can promote more positive perceptions of AI in learning, informing AI Literacy programs in K 12 settings. Limitations include the lack of AI-specific teacher readiness measures, no Assessment of school infrastructure or students' actual AI access/use, and the absence of fine-grained teacher technological-knowledge indicators.

Connected Concepts

Connected Articles

Citation

Baez, C., Buchner, J., Ullrich, A.-L., & Schallert-Vallaster, S. (2026). Exploring AI perceptions in education: unveiling the role of student and teacher motivation and self-efficacy. Computers and Education Open, 100346. https://doi.org/10.1016/j.caeo.2026.100346