Artificial intelligence tools are already in the hands of Nigerian students. A Lagos educator is now raising a question that deserves more attention than it is getting: what happens when students use AI to answer questions they do not have enough knowledge to verify?
Wilfred Dieke, Director of Helmbridge Study Centre, warned that inadequate understanding of a subject could make it difficult for students to identify inaccurate information generated by AI tools, stressing that AI should be regarded as a tool that complements learning rather than a substitute for the foundational knowledge required to use it effectively.
Dieke made the remarks at the inaugural ceremony marking Helmbridge Study Centre’s 50th anniversary in Lagos, attended by academics, legal professionals, alumni, and parents.
Dieke used a simple analogy to make his point, one that any Nigerian parent or teacher will immediately recognise.
He said that just as a child is introduced to a calculator, you do not give a primary school child who has not understood the basics of arithmetic a calculator to start off. You want that child to get mastery over basic arithmetic and quantitative reasoning before using those tools to fast-track information.
The parallel to AI is direct. A student who does not understand how an argument is constructed cannot tell when an AI has constructed a bad one. A student who has not studied a topic cannot tell when an AI has filled gaps in its answer with invented information.
Dieke specifically warned about AI hallucinations, saying that if students do not have a working knowledge of where they are going, it gets murky, and AI is able to hallucinate. Without enough background knowledge, a student cannot detect when there is hallucination in the information being given.
AI hallucination is the term used when a language model generates information that sounds confident and coherent but is factually wrong or entirely invented. It is one of the most dangerous characteristics of current AI tools for educational use, precisely because the output looks identical to accurate information. A student who submits an AI-generated answer containing a hallucinated fact, statistic, or source has no way of knowing it is wrong unless they already know enough about the subject to check it.
Dieke added that students needed to be consciously taught where and how to use AI, particularly because they were already being exposed to the technology regardless of whether schools officially introduced it or not.
This last point is important. Nigerian students are already using ChatGPT, Gemini, and other AI tools for assignments, research, and study. The question is not whether to introduce AI into education. It is whether educators are giving students the critical framework to use it safely.
Dieke’s argument is not that AI has no place in education. It is that the sequence matters. Foundation first, tools second.
This is a position that is gaining ground among educators globally as the initial excitement around AI in classrooms gives way to harder questions about what students are actually learning when they outsource thinking to a machine. Research from several universities has found that heavy AI use without sufficient foundation knowledge leads to students who can generate plausible-sounding answers but struggle to explain or defend them under examination conditions.
For Nigerian students preparing for WAEC, JAMB, post-graduate professional exams, or competitive university entry, the ability to reason through a problem independently is still what gets tested. AI can help with revision, summarising content, and generating practice questions, but it cannot sit the exam for the student. The foundation still has to be there.
The practical message from Dieke’s warning is straightforward. Before encouraging or allowing a child to use AI tools for schoolwork, make sure they have enough grounding in the subject to tell a good answer from a bad one.
For schools, it means AI literacy cannot be limited to showing students how to write a prompt. It must include teaching them how to critically evaluate what comes back, cross-check claims against reliable sources, and recognise the limits of what AI actually knows versus what it confidently presents as fact.
Dieke’s broader point was that the same principles Helmbridge has applied for 50 years, building character, discipline, and intellectual foundation before layering on tools and techniques, apply just as much in the age of AI as they did before it existed.
The tool changes. The need to understand what you are doing with it does not.
