Stanford Study: Offering AI Tutoring Tools Alone Falls Short on Engagement and Reading Gains

A Stanford University study released Wednesday found that merely offering AI tutoring tools to students is not enough to create meaningful engagement and improve reading outcomes. In two school districts, only about 61% and 53% of students used the AI tutor even when time was scheduled, with average weekly usage of 2.18 minutes and 5.23 minutes respectively. When the AI tutoring platform was used alongside human tutors, engagement rose by only 1 minute and 4.4 minutes per week, still far below the 30 minutes of weekly use recommended by the platform provider.

The findings give schools and policymakers implementation-level evidence as they decide whether to adopt AI tutoring platforms: personalized instruction depends on consistent student use, and offering tools or scheduled time alone does not guarantee engagement. The study also suggests that engagement differences may affect AI tutoring's role in reading outcomes, while state pilots have already reported lower-than-expected usage. The sample covers specific grades and tutoring arrangements in two districts, and no independent quantified reading-effect conclusion was provided, so the results cannot be used to claim AI tutoring is broadly effective or ineffective. Schools evaluating AI tutoring platforms need to include actual student usage time and engagement mechanisms in their assessments, not just tool features. Policymakers designing AI tutoring pilots may need accompanying training, scheduling and oversight to narrow the gap between recommended and actual usage.