All guidesThe Learning Twin
The Learning Twin
AI student: A private model of what you know — and you can read all of it.
Your Learning Twin is a per-course model of you as a learner: how well you know each concept, what you currently misunderstand, and what to study next. It grows quietly as you work — and everything it holds is visible to you.
What it is
Three things live in the twin: a mastery score per concept (updated with every quiz answer and lesson), a set of open misconceptions — specific misunderstandings, paraphrased, never quotes of your words — and a weekly study plan built from your weakest concepts and the lessons you have left.
It also writes a short narrative digest: what you have been working on, where you are winning, and what to focus on this week.
Why — the pedagogical rationale
Learners cannot self-assess accurately — we are systematically over- or under-confident about specific concepts while feeling generally fine. A twin gives you an honest mirror at concept granularity, not a vague percentage.
Transparency is the point, not a bonus: a model of you that you cannot inspect is untrustworthy. Your twin shows you every mastery signal, every misconception, and every plan decision — and you can challenge any of it.
How we achieve it
Quiz grading, lesson completion, and tutor conversations each fire a signal into the twin. Quiz answers update an Elo-lite mastery score per concept — big moves early, smaller ones as evidence accumulates, with a gentle decay toward neutral when a concept goes untouched for weeks.
A periodically-run synthesis pass turns your recent activity into the narrative digest and misconceptions, throttled so it costs little. The weekly plan is generated on demand from what the twin knows plus the published lessons you have not finished.
Where to find it
The floating twin launcher on any course page opens your twin panel; the full page lives at /twin. Your teacher sees class-level statistics only — never your individual twin.