AfriversalAI's proprietary decision framework. Five questions. Any AI situation at work. Approximately two minutes. No technical knowledge required.
The Funda Five is the thread that runs through every module, every sector track, and every professional cohort. It is introduced in Module 1 and applied every subsequent module. By the end of the course, using it is instinctive.
Not every task suits AI assistance. This step assesses the risk profile of the task, what happens if the AI output is wrong, and whether the consequences are reversible. A draft email is low risk. A public policy citation is not.
What data is being fed into or produced by this AI system? Is it confidential? Is it personal information covered by POPIA? Is the AI tool authorised to process this type of data? Learners leave this step with a data sensitivity instinct.
Different AI tools have different failure modes, risk profiles, and organisational approval status. Is ChatGPT approved for this task? Is this the right type of AI — generative, predictive, computer vision? Learners develop tool awareness, not tool loyalty.
Before acting on AI output, can you articulate a specific, achievable verification step? "I'll cross-check the citations" is a verification step. "I'll read it carefully" is not. This step turns verification from a vague intention into a concrete action.
"The algorithm decided" is not a legal or ethical defence under South African law. This step requires learners to name the accountability chain — a specific person or role responsible for the decision — before AI output reaches anyone who could be affected by it.