Glocalizing Generative AI in Education for the Global South: The Design Case of 21st Century Teacher Educator AI for Ghana

Localization matters for AI equity in the Global South: The tool's integration of Ghana's National Teacher Education Curriculum Framework, local languages (Twi, Dagbani, Mampruli, Dagaare), and culturally responsive pedagogies demonstrated that customized GenAI systems can provide more relevant support than generic global models, addressing a key equity gap for pre-service teachers in resource-constrained contexts.

Responsible AI requires intentional design against neocolonialism: The researchers explicitly incorporated frameworks addressing "GenAI neocolonialism" and used UNESCO's AI guidelines to ensure the tool served local educational priorities rather than imposing external agendas, suggesting that responsible AI deployment in developing regions demands proactive safeguards against cultural and educational dominance.

Structured data and human oversight improve reliability: By using preloaded, curated datasets (rather than open web scraping) and an Interactive Semi-Automated Prompting Strategy requiring user input about institution and semester, the design reduced hallucinations and ensured outputs aligned with Ghana's actual curriculum—demonstrating that responsible GenAI requires deliberate constraints, not maximum autonomy.

Bilingual and contextual support enhances learning outcomes: Pilot results showed the tool's strength in delivering bilingual support and localized examples, suggesting that GenAI tools designed with linguistic and cultural adaptation can improve comprehension and engagement for learners whose needs are typically underserved by English-only, globally generic AI systems.