Resources
Decolonial Prompt Corpus & Open Lesson Bank
From Critique to Creation
The Participatory Action Research journey does not end with critiquing global AI; it culminates in the creation of sovereign, functional educational tools. By transforming our cohort of Ugandan educators from passive consumers into active co-designers, the RGAI-FCT project has produced a suite of ethical, culturally anchored AI support tools tailored specifically for fragile and resource-constrained learning ecosystems.
The Open Lesson Bank
A Community-Owned Repository
The Open Lesson Bank is a dynamic, searchable database of AI-enhanced Open Educational Resources (OERs) and classroom activities co-created entirely by Ugandan teachers operating in Accelerated Education Programmes (AEP) and refugee settlements. Rather than relying on generic, Western-centric AI outputs, educators across the continent can access lesson plans that are explicitly designed to respect African epistemologies, community values (Ubuntu), and local infrastructure.
The database is categorized into three core discipline tracks:
Track 1: Sciences
STEM, Health, Engineering, Environmental Sciences
- AI-supported activities modeling local climate resilience
- Agricultural productivity and health innovations
- Protection of indigenous agroecological knowledge
Track 2: Humanities
History, Social Sciences, Philosophy, Arts
- Critique colonial archives using AI
- Summarize historical datasets
- Reconstruct marginalized African narratives
Track 3: Languages
English, African Languages, Literature
- Re-engineer AI translation tools
- Evaluate linguistic accuracy
- Preserve cultural nuance of African literature
The Decolonial Prompt Library
Reclaiming Epistemic Control
Because global Large Language Models (LLMs) default to their Western training data, a generic prompt will inevitably produce a generic, Eurocentric outcome. The Decolonial Prompt Library institutionalizes the pedagogical wisdom of Ugandan teachers by compiling highly engineered, peer-reviewed prompts based on the RTCC Framework (Role, Task, Context, Constraints).
See the Difference: Epistemic Audit in Action
Raw AI Output
"Write a science lesson on the water cycle."
The AI generates a digital-heavy lesson plan featuring examples of snowmelt, four-season temperate climates, and online interactive quizzes. It completely ignores the realities of the Ugandan classroom.
Engineered RTCC Prompt
[Role] Act as an expert Ugandan secondary school teacher specializing in the Competency-Based Curriculum (CBC). [Task] Create a 40-minute lesson plan on the water cycle. [Context] You are teaching 60 learners in an AEP refugee settlement. You have no internet access, no digital devices for students, and must use oral storytelling and locally available materials. [Constraints] You MUST align the competencies with the National Curriculum Development Centre (NCDC) syllabus. DO NOT use examples of winter or snow; strictly use examples relevant to Uganda's wet and dry seasons. Include a formative assessment that does not require printed paper.
A highly practical, culturally resonant lesson plan that respects the infrastructural realities of the classroom, honors local environmental contexts, and perfectly aligns with national educational standards.
Browse the Decolonial Prompt Library
The Localized Custom LLM
African AI Sovereignty
While prompt engineering mitigates the bias of foreign platforms, true digital sovereignty requires owning the infrastructure. The culmination of the RGAI-FCT project is the development of a bespoke, fine-tuned Large Language Model (LLM) built specifically for Ugandan schools and fragile contexts.
Trained on Safe, Localized Data
Unlike global models that scrape the internet indiscriminately, our custom LLM is fine-tuned using a secure, meticulously curated dataset. This includes the RTCC prompts generated by our teachers, NCDC-aligned curriculum materials, and local language datasets.
Ethical and Non-Extractive
Built on the principles of Decolonial AI, this model ensures that the data provided by African institutions is not extracted for foreign corporate profit. The data is processed securely, often utilizing offline or low-bandwidth capabilities.
A Sovereign Teaching Partner
By embedding the pedagogical reasoning, cultural references, and safety guardrails defined by the teachers themselves, this localized LLM serves as a reliable, context-aware teaching assistant that amplifies human dignity rather than eroding it.
Start Using These Tools Today
Access our growing repository of culturally anchored, AI-integrated resources.