AI Helps Teachers Create Culturally Relevant Lessons

In Ghana, as in many parts of the Global South, STEAM education often feels disconnected from students' daily lives. Science, technology, engineering, arts, and mathematics are frequently taught using Western examples and abstract concepts that fail to resonate with local realities. This gap between curriculum and culture has long been a for educators striving to make learning meaningful. A new study from researchers at the University of Georgia and several Ghanaian institutions offers a potential solution: generative artificial intelligence, when carefully guided, can help create lesson plans that bridge this divide.

The research examined how generative AI-produced lesson plans compare to standardized curriculum materials in supporting culturally responsive teaching. Four STEAM education experts with deep knowledge of Ghanaian classrooms evaluated both AI-generated lessons and Ghana National Council for Curriculum and Assessment (NaCCA) lesson plans across mathematics, science, creative arts, and computing subjects. They used a validated 25-item rubric assessing five domains: bias awareness, cultural representation, contextual relevance, linguistic responsiveness, and teacher agency. The AI-generated lessons were created using a customized Culturally Responsive Lesson Planner (CRLP) tool with interactive semi-automated prompts designed to incorporate local languages, practices, and examples.

Revealed a clear pattern: AI-generated lessons outperformed standard curriculum materials in several key areas. Teacher agency emerged as the strongest domain, with a mean score of 3.6 out of 4, indicating that the CRLP tool effectively supported teachers in creative instructional design and customization. Contextual relevance also scored highly at 3.4, showing that AI could successfully integrate local realities, traditional knowledge, and community-based examples into lesson plans. For instance, in computing lessons about information security, the AI-generated plan used examples from mobile money transactions (MoMo) familiar to Ghanaian students, while science lessons connected water cycles to local farming practices for crops like maize and yam.

However, the study also identified significant limitations. Cultural representation was the weakest domain, scoring only 2.25, with particular gaps in mathematics and computing where AI struggled to authentically capture cultural nuance and diverse identity representation. Linguistic responsiveness scored moderately at 3.05, with experts noting that while the AI incorporated local languages like Dagbani and Dagaare, translations were sometimes inaccurate and some learners might still be excluded if they didn't understand the specific local language used. The researchers found that AI-generated lessons often included surface-level cultural references without deeper understanding of Ghana's cultural pluralism.

For education in Ghana and similar contexts are substantial. The study demonstrates that AI can help address the persistent problem of Westernized, decontextualized STEAM education by connecting abstract curriculum standards to learners' sociocultural worlds. The CRLP-generated lessons integrated indigenous knowledge, bilingual elements, community artifacts, and locally grounded analogies that made learning more relevant and engaging. This approach could significantly reduce teacher workload while improving educational quality, as educators spend less time creating materials from scratch and more time adapting AI-generated content to their specific classroom needs.

Despite these promising , the researchers emphasize that AI cannot replace human judgment and cultural expertise. The study's limitations include its small sample size of four experts, focus on lesson plans rather than classroom implementation, and reliance on AI models not fine-tuned on local Ghanaian corpora. Future research should examine how these AI-generated lessons perform in actual classrooms, involve more diverse experts including practicing teachers, and explore model fine-tuning using indigenous language datasets to improve cultural fidelity. The path forward requires thoughtful partnership between human insight and technological innovation, ensuring AI becomes a tool for educational equity rather than another source of cultural homogenization.