Course insights
For educators
Students studying with Alexandria ask questions in their own words, at the point of confusion, against a specific piece of material. An assessment records who misunderstood something after the fact; this records what was unclear while the course is still running.
The activity tab provides four views.
Dashboard
Section titled “Dashboard”The headline numbers are total questions, active students, and questions per student. More useful is the breakdown underneath, which tells you how students are working:
| Type | What it means |
|---|---|
| Sourced | Answered directly from your course materials |
| Socratic | The AI guided with follow-up questions instead of answering |
| Text Highlight | Asked about a passage they highlighted |
| Image Highlight | Asked about a figure or diagram they highlighted |
| Unsourced | Answered without source checks |
| Web Search | Went beyond the course |
| Custom Chatbot | Used a chatbot you configured |
| Exam Tutor | The practice-exam tutor flow |
A high Image Highlight count on a mathematical course generally indicates that diagrams carry more of the explanatory load than the text. A high Web Search count suggests the course assumes background it does not itself supply.
Student engagement distinguishes returning students, active students, and never active. A large never-active figure indicates a problem of access or awareness rather than of teaching, and will not be resolved by changes to the course.
The other three views
Section titled “The other three views”Each has its own page:
- The heatmap: which page of which document the AI cites most, and the questions that drew on it
- Reading conversations: the exchanges themselves, read-only
- Course reports: weekly written summaries with recommendations, and an all-time course report
Interpretation
Section titled “Interpretation”The informative signal is not the busiest topic but the topic that is busier than expected.
Heavy questioning of material known to be difficult is confirmation. Heavy questioning of material assumed to be straightforward is the finding, and warrants attention in the following session.
Volume alone does not indicate a problem: sustained questioning of material also indicates engagement with it.
Privacy
Section titled “Privacy”Insights are teaching signal, not surveillance.
The purpose is to identify where a cohort is struggling, not to monitor individuals. Stating plainly to students that topic-level question data is visible is advisable, and tends to increase rather than suppress questioning, since it presents asking as a contribution.