<![CDATA[The rapid development of generative artificial intelligence GenAI brings opportunities to K-12 education while simultaneously exacerbating the digital divide in low-resource regions. This study addresses the realities of backward infrastructure and unstable networks in Chinese rural high schools by designing and validating a fully localized, zero-cost offline GenAI maker education framework. It utilizes the Ollama offline engine, Open WebUI interface, and text adventure game templates to achieve multimodal interaction in network-free environments. Centered on Seymour Papert’s constructionism theory, it proposes the “Offline GenAI Maker Education Peak Progressive Framework.” Through an 8-week progressive curriculum basic narrative → cultural multimodal → emotional companionship → ethical guardianship → comprehensive originality → meta-reflection, a quasi-experimental intervention was conducted in a Grade 10 class N=50 at a rural high school in Hebei Province. The results show that students’ AI literacy exhibited significant multi-stage peak evolution: rapid early increases in textual depth and cultural integration, successive mid-term peaks in emotional engagement and ethical critique, later expansion of originality capabilities, and, by Week 8, the achievement of closed-loop internalization of cognition–emotion–responsibility through meta-reflection. All dimensions showed pre- and post-intervention effect sizes of Cohen’s d > 3.2 p < 0.001, using a researcher-developed 5-point standardized scale; due to the extremely low baseline among rural students and the narrow scoring range, a certain ceiling effect exists. All 50 students produced complete original interactive games that met the preset evaluation standards. The study validates the localized pathway of constructionism in low-resource contexts in the digital era, fills the empirical gap in fully offline GenAI education for rural high schools, and provides a replicable Chinese solution for educational equity in the Global South]]>
Copyrights © 2026
