How we check our past questions
Past questions are only useful if the answer key is right. This is how we look for wrong keys and what we do when we find one.
Key-check rounds
We check the question bank in rounds. In each round, AI solvers from more than one provider answer the questions without seeing our answer key. When they all reach the same answer, with high confidence, and it differs from the key, that is strong evidence the key is wrong. Questions that need a diagram or a passage the solver cannot see, and subjects where an AI solver is not reliable enough (such as languages and literature), are never changed automatically. They go to a person.
People decide the hard cases
Every disagreement that is not clear-cut goes into a review queue in our admin tools, where a person reads the question and settles the key. When an admin confirms a key, later automatic checks do not overwrite it. Each change keeps the previous version of the question on record, so a fix can be undone.
How AI explanations are used
An AI explanation is shown to everyone only when it reaches the same answer as the answer key. If an AI solution disagrees with the key, it is not shared. Instead the disagreement is flagged for a person to look at. If an explanation turns out to be wrong, we can switch it off, and it is replaced when the question is edited. AI-written explanations are labelled as AI explanations; they are never passed off as the official explanation.
Students can report a question
If an answer or an explanation looks wrong, use the report option on the question in the app, or write to us through the contact page. Reports go to the same review queue.
What this does not mean
Checking reduces errors; it does not remove them. Past questions come from past exam papers, and we are an independent study app: we are not affiliated with JAMB, WAEC or NECO, and the exam bodies’ own publications are the authority. Read more in our editorial policy.