One Click Away: AI Tutoring with Khanmigo in a Two-Year School Experiment
Summary
The article presents early evidence from a large-scale trial of Khanmigo AI tutoring in K-12 math, showing modest achievement gains linked to active use. It finds engagement to be the primary bottleneck, with most students attempting Khanmigo infrequently and often not engaging in substantive dialogue. The takeaway is that simply providing AI tutoring access is insufficient; effective benefits depend on user engagement and how the tutor is integrated into practice.