Exploring Vocational Teachers’ Perceptions of Generative AI for Teaching-Material Development: A Pilot Study

Abstract

This pilot study examines how vocational teachers’ self-assessed competency in using generative artificial intelligence (GenAI) is associated with perceived productivity, perceived teaching-material quality, and confidence in continued GenAI use. Eighteen productive-subject teachers at one State Vocational High School in Bandung completed a post-training questionnaire; one non-substantive test response containing multiple placeholder fields was removed, leaving 17 valid cases. The analysis combined descriptive statistics, descriptive categorization of open-ended responses, an exploratory partial least squares structural equation model (PLS-SEM), and sensitivity checks. The sequential model linked self-assessed AI Competency (AIC) to Perceived Productivity (PRO), PRO to Perceived Quality (QUA), and QUA to Continued-Use Confidence (CUC). Path coefficients were positive (β = 0.810, 0.958, and 0.942), with bootstrap p < 0.001 for all paths; R2 values were 0.656, 0.918, and 0.888. However, HTMT between PRO and QUA was 0.982, so the PRO–QUA relationship is not interpreted as a distinct construct-level effect despite its statistical significance. Rank-based and leave-one-out sensitivity checks showed the same directional pattern. Infrastructure barriers remained salient descriptively (Q10: M = 3.65, SD = 1.00). The findings provide exploratory single-school evidence, not generalizable causal conclusions, and emphasize competency development, human validation of AI output, institutional support, and stronger future measurement design.