Solution
Three ways of learning that lock together
Memorising facts about AI doesn't make a student more resistant to it. What builds that resistance is answering, being wrong, reading why, and then arguing with someone else about a case with no single right answer — which is what these three parts of Minervas are built to do.
0
Published questions
0
Forum threads
0
Forum replies
0
Students learning
A quiz that explains rather than punishes
Every question is followed by an explanation shown after the student answers — including when they're right, because right-by-guess and right-by-understanding are not the same thing, and the explanation is what tells them apart.
The answer key is never sent to the browser. Scoring happens server-side through a Postgres RPC, so a score reflects what a student actually knew, not what could be read out of the page source.
An AI that asks back
AIIn the forum, the AI withholds its conclusion until the student has argued their position across several turns — one question, under 40 words, every reply. It chases the student's reasoning instead of handing them an answer.
Everything the AI writes carries the same colour and label everywhere in the app, drawn from a database column, never from a display name that could be spoofed. In a product that teaches people to recognise machine writing, letting machine writing pass as a person's would be an expensive irony.
Points that are hard to game
Points can only be minted by the database, never by a request from a browser. Re-answering the same question adds nothing, and forum upvotes deliberately mint no points at all — so two accounts can't lift each other's scores.
Teacher and student roles are fixed at signup, and every forum reply shows a Student/Teacher/AI badge pulled from that same database column — never from a name a user could type in themselves. What a reply is worth is decided by what it says, not by the badge next to it.