FROM USERS TO CREATORS: BUILDING A LOCAL EDUCATIONAL AI ECOSYSTEM BASED ON INDONESIAN LANGUAGE AND CULTURE

The global proliferation of Generative AI (GenAI) presents a risk of technological and epistemic dependency for non-Anglophone nations like Indonesia. Current models, trained on Western data, exhibit significant linguistic and cultural misalignment, failing to represent Indonesian national concepts (e.g., Pancasila) or local languages (bahasa daerah). This misalignment undermines national education goals. This research aims to conceptualize and propose a comprehensive framework for a sovereign Indonesian educational AI ecosystem, initiating a strategic pivot “from users to creators.” A constructive research methodology was employed, synthesizing a systematic literature review (N=120), national policy document analysis (N=25), and technical benchmarking of state-of-the-art models. The results identified a “cycle of dependency” characterized by three findings: (1) a dominant “User Paradigm” in national academic research (95% focus on adoption); (2) a critical “policy integration gap” between siloed government ministries; and (3) definitive technical-cultural misalignment of global AI models, which failed to process core national concepts. The study concludes by proposing the Sovereign Indonesian Educational AI Ecosystem framework as a novel, constructive artifact. This framework provides an integrated strategy to break the dependency cycle, advocating for a “decolonized AI” approach centered on developing sovereign models from a curated National Language and Culture Corpus