Why every AI sentence here tells you how much to trust it
Four labels sit on every AI explanation on Meyshan Med — and the strictest one wins by default. Here's the honesty machinery.
There's a question every medical student should ask any AI study tool, and almost none of them answer: how do you know that?
On Meyshan Med, every AI-generated explanation carries one of four labels, always visible, never buried in a footer:
Directly supported — the sentence is grounded in the material you supplied. Your chapter, your notes, your upload. The strongest label, because the source is sitting right there.
Simplified by Meyshan — a human medical student reviewed this simplification and stands behind it. A name, not a model, takes responsibility.
Additional context — the AI added something useful beyond your material. Probably right, honestly labelled as added, so you know to check it the way you'd check a senior's offhand comment.
Medical review required — the system isn't sure enough to stand behind this yet. It says so, out loud, instead of pretending.
The detail I'm proudest of is the default: when the AI fails to label its own sentence, the system doesn't guess generously. It assigns Medical review required — the strictest label. Uncertainty rounds down, never up. Marketing would round up. Medicine rounds down.
The same principle runs through everything here. The Misconception Radar ships with sixty classic traps written and waiting — and every single one is switched off until Meyshan personally reviews it, because a quiz question that's wrong is worse than no question. New Question Bank items pass automated structure checks, then an AI answers them blind, then an AI attacks them — and none of that certifies anything, because the advisors advise and only a human medical review publishes. The rule is structural: content literally cannot flip itself to "reviewed."
Why be this strict on a free student project? Because the whole point of studying is building a foundation you can stand on in front of a patient someday. A study tool that quietly mixes verified and unverified sentences is teaching you a habit that gets dangerous later. The labels aren't a legal disclaimer — they're the curriculum.
So when you read an explanation here and see the small label on it, that's the system telling you, honestly, how much weight it can bear. It's the feature I'd want every study AI to steal.
— Meyshan Med · The full review methodology · See it in the app