What's this about?
A two-year cluster-randomized study in 18 middle schools found small maths gains from access to Khanmigo, comparable to ordinary Khan Academy practice. Students tried it, but rarely used it for sustained mathematical dialogue.
What exactly is a cluster-randomized trial?
Whole groups such as schools or classes, rather than individual pupils, are assigned to conditions. This reduces spillover but leaves fewer independent units to compare.
What's the community saying?
HN 42 points, 30 comments at selection time.
- Readers focused on the conclusion that use, not access, was the binding constraint.
- Some were surprised AI tutoring did not clearly beat static study materials.
- Parents and educators noted that different children respond very differently to educational software.