<![CDATA[Background: The educational tension in vocational schools for tourism in Bali is specific. The digitalisation of assessment often conflicts with the preservation of locally valued service competencies embedded in the Balinese Hindu pedagogy of Samskara (ingrained habits), Cesta (volitional effort) and Kriya (embodied skills). At present, no validated assessment tool can measure all three constructs simultaneously. Aims: This study develops and validates an AI-driven e-assessment instrument for them among Balinese tourism VET students. The novelty lies in operationalising culturally embedded competencies rather than generic digital skills. Methods: It utilised a sequential mixed-methods design. Phase 1: Conduct ten semi-structured interviews with experts on culture and education to generate measurable indicators. Phase 2: Development of a 48-item AI platform for assessment using natural language processing, affective computing and computer vision. Phase 3: Pilot with 320 Grade XI learners from three schools with varying degrees of digital infrastructure. Result: Internal consistency was acceptable (Cronbach’s α = 0.89, 0.87, 0.91). Confirmatory factor analysis supported the three-factor structure (CFI=0.94, RMSEA=0.058). Across all three competences, students in moderately digitised schools outperformed students in low and highly digitised schools (p < .01, Cohen’s d = 1.46–2.48). The quadratic regression analysis showed a negative quadratic coefficient which suggests an inverted-U relationship or the best heritage competency outcomes. Conclusion: AI-driven assessment can validly measure culture-specific competencies in tourism VET. The inverted-U pattern supports the notion of a “smart tourism heritage tension” rather than confirming a paradox. Caution is warranted regarding generalizability, cross-sectional design, and potential algorithmic bias.]]>
