A Tennessee district's math software trial shows gains came from the platform, not the AI tutor
A two-year randomized study of 18 middle schools found faster math gains from Khan Academy even though students barely used its bundled AI chatbot.
This is not our reporting
This story was reported and published by Chalkbeat. What follows is our summary of it, and why it matters for people writing policy.
Read the original at Chalkbeat ↗Our summary
According to the piece, a two-year randomized study in a Tennessee school district assigned low-performing middle schoolers either to Khan Academy's self-paced online math program, bundled with the AI chatbot Khanmigo, or to a mix of other digital tools and teacher-led small groups. Students using Khan Academy made faster math gains, but researchers found the improvement came from the platform itself, not the AI: students engaged with Khanmigo only about a third of the days, often sending off-topic messages rather than seeking help. For districts considering AI tutoring add-ons, the finding suggests structured, self-paced practice mattered more than the chatbot layered on top.
Read the full story at Chalkbeat ↗
More from The Effective
Tulsa schools flag eviction filings so families get help before kids fall behind
A Tulsa Public Schools partnership with a local nonprofit alerts parents facing eviction to legal aid and rental assistance, aiming to keep children enrolled.
An Indianapolis school pairs reading intervention with housing help for families
A nonprofit-run Indianapolis Public Schools campus rebuilt around family support and daily reading blocks is showing early gains after years as one of the district's lowest performers.
Philadelphia cut school weapon confiscations by two-thirds through relationship-building, not just security
District data show weapons confiscations down 67% and fights down 22% since 2021, which the safety chief credits to conflict mediation and trusted-adult networks rather than added policing.