Conversational AI agent for teacher education in Ghana

While generic and frontier Artificial Intelligence (AI) tools are expanding rapidly in global education, their lack of contextual sensitive and cultural awarenenss limits their efficacy in localized settings. Consequently, limited attention has been given to developing context-aware, culturally grounded, and region-specific AI systems for teacher education. This study is among the first to examine the development and evaluation of GenAITEd Ghana, a context-aware and curriculum-aligned conversational AI prototype designed for Ghana teacher education. Using a design science research approach, we developed a multi-agent and retrieval-augmented AI system grounded in Ghana’s National Initial Teacher Education curriculum. We evaluated it through functional feasibility testing, comparative alignment with Responsible AI frameworks, and reflexive thematic analysis of feedback from teacher education experts. The feasibility testing showed that the GenAITEd Ghana AI tool functioned reliably across its core components, including teacher and student login by institution, semester and specialization, curriculum alignment, subject-specific AI agent customization, group creation through access-code-based entry, multilingual text and voice interaction, and teacher oversight of students' AI-supported learning spaces. It also included teacher voice cloning and avatars, although these remained developmental at the time of evaluation. Simultanously, the evaluation demonstrated that the GenAITEd Ghana aligns with key Responsible AI dimensions such as transparency, fairness and inclusion, privacy and security, accountability and governance, and human-centered pedagogy. In addition, the expert review suggested that GenAITEd Ghana was pedagogically relevant, culturally grounded, and closely aligned with the Ghanaian teacher education context. Nonetheless, the experts called for more profound cultural embeddedness, and expressed concerns about teacher educators’ technical expertise to customize their subject-specific agents and manage student groups effectively. This study highlights the promise of GenAITEd Ghana as a locally governed AI tool for Ghanaian teacher education. We therefore call on the Ministry of Education, Ghana, to collaborate with appropriate stakeholders to strengthen, scale, and institutionalize GenAITEd Ghana in the country. The study further positions GenAITEd Ghana as a model that may inform locally governed AI systems in other Global South contexts.