Why Students Fall Behind Before Assessment Results Reveal It

University educators often have limited visibility into which students have really understood a concept during a class. A few answer questions, some look engaged, and many stay silent. When instructors assume that silence means understanding, a structural blind spot emerges. If gaps are not addressed quickly, they accumulate, making later topics harder to master.

Writing in Times Higher Education, Zabin Visram argues that just-in-time support is the answer. Rather than waiting for end-of-semester assessments, lecturers could use AI to create frequent, low-stakes checkpoints. Short AI-generated diagnostic quizzes can assess understanding after each topic or bi-weekly, and the results can route students into individual learning pathways.

Those pathways might include brief readings, case studies, two-minute explainer videos or podcasts. In the author's own teaching, Visram says, the approach brought previously disengaged students back into the learning process. Students described the short materials as manageable, less overwhelming and easier to fit into a busy schedule.

The model outlined has five areas of action: checkpoint insight, pedagogy, modular content, differentiated learning pathways and immediate feedback. AI tools such as Copilot, ChatGPT and Claude can create materials from lecture notes, while Synthesia and HeyGen can turn written content into video. The approach does not require advanced systems, but it does place new demands on instructors and institutions.

What Just-in-Time Adaptive Learning Actually Changes

The blind spot is built into conventional course design

The core problem is that teaching often assumes understanding unless a student signals otherwise. Because only a few students speak up, large groups can move on while fundamental gaps remain. End-of-course assessment confirms learning too late to allow simple correction.

What AI tools such as ChatGPT, Claude and Synthesia change

The practical shift is a reduction in the time needed to generate differentiated materials. A lecturer can turn lecture notes and selected readings into quizzes, short case studies, readings or video scripts within minutes. That lowers the barrier to offering multiple pathways: advanced learners can be challenged while emerging learners consolidate the same topic at their own pace.

The workload does not disappear; it moves

The model adds new tasks for instructors: designing checkpoints, selecting AI tools and reviewing the generated content. Visram is explicit that adaptive learning requires new skills and time. Institutions that do not provide training, tools and workload recognition risk shifting the effort from assessment outcomes to an uncounted preparation burden.

Why this remains a promising practice rather than proven evidence

The reported student re-engagement comes from one instructor's teaching experience, not from a controlled study. The absence of independent outcome data means universities should treat the just-in-time adaptive learning model as a practical starting point to test, not an established benchmark. The strength of the argument is its specificity; its limitation is that personal testimony cannot prove scale.

Practical Steps for Lecturers and Universities Adopting the Model

For lecturers, the model translates into a small set of concrete practices. The bullets below follow directly from the five areas Visram describes.

  • Build AI-generated, low-stakes diagnostic quizzes after each topic or bi-weekly. Their purpose is to reveal gaps before the next learning block begins, not to add summative pressure.
  • Route students by quiz score into differentiated pathways: recovery resources for emerging learners and more advanced material for proficient students.
  • Use Copilot, ChatGPT or Claude to turn lecture notes and selected readings into short readings, case studies, two-minute explainer videos or podcasts.
  • Use Synthesia or HeyGen when text needs to become video, and use speech-to-text tools or assistants such as Google Assistant and Apple Siri to support planning and summarising.
  • Use AI agents to give individual, incremental feedback on problem-solving tasks; the author found this feedback student-friendly and interactive.
  • If you are a university leader, treat this as workload, not free capacity: provide training, approved tools and recognition for the extra time the approach initially requires.