STUDYLABPRO
RU

All research

Human–Machine Learning

Adaptive Mastery Models

STATE / Active — in production use

Hypothesis

A student who sees exactly where they are stuck, and receives the smallest next step that produces a win, recovers their learning trajectory — motivation is born from success, not gifted.

Problem

Linear courses assume nobody falls behind. Everybody falls behind. The interesting problem is diagnosis and recovery, not content delivery.

Current state

Active — in production use.

Methods

  • Mastery estimation over a subject knowledge graph
  • Gap diagnostics: locating the failed prerequisite, not the failed exercise
  • Next-step selection tuned for visible small wins

Findings

  • Gap-first sequencing changes learner behaviour within days, not months
  • Mastery must be modelled on the graph, not on the exercise list