<![CDATA[Indonesia’s Coding and Artificial Intelligence (KKA) curriculum mandate (BSKAP No. 046/H/KR/2025) requires Vocational High School (VHS) students to master Python programming and generative AI, exposing infrastructure inequality and cognitive overload as dual barriers. This study reports five ADDIE phases of developing and evaluating a cognitive scaffolding module integrated with Vilokka, an on-premises virtual laboratory embedding a guardrailed Local Large Language Model (Local-LLM). Participants were recruited through purposive sampling: five KKA teachers (Analysis), three validators (Expert Validation), and 30 VHS students (Implementation and Evaluation). Analysis identified three cognitive barrier patterns: pre-coding abstraction failure, logic-syntax conflation, and AI-induced cognitive passivity. The Design phase produced a 17-module architecture synchronising a three-level scaffolding framework (Sense-making, Process-scaffolding, and Fading) with KKA curriculum elements. The study employed four instruments: a teacher interview guide, an expert validation rubric, pre-test and post-test assessments, and practicality questionnaires. Expert validation yielded a mean validity of 89.72% (Very Valid). The field trial yielded a mean N-Gain of 0.39 (Moderate), confirmed by paired sample t-test (t(29) = 6.385, p < 0.05; Cohen’s d = 1.17). Practicality ratings were Very Practical for teachers (88.12%) and students (86.96%). The findings establish a replicable framework for AI-assisted vocational programming instruction, recommended for schools with infrastructure constraints.]]>
Copyrights © 2026
