AI Tutors Fail to Spark Widespread Learning Gains

Chalkbeat reports (April 2026, by Matt Barnum) that several high-profile predictions about AI tutors have not produced a broad transformation in student outcomes. The article cites a 2023 statement from Sal Khan calling AI tutoring "probably the biggest positive transformation that education has ever seen," and a 2024 remark from OpenAI CEO Sam Altman about "virtual tutors who can provide personalized instruction in any subject, in any language, and at whatever pace they need," both cited by Chalkbeat. Chalkbeat reports that access to Khan Academy's AI tutor Khanmigo grew from 40,000 students in 2023 to nearly 1 million in 2026, yet uptake and regular use have stagnated. Chalkbeat quotes Kristen DiCerbo, Khan Academy's chief learning officer, on student engagement challenges. A separate Stanford University study (June 2026, reported by K-12 Dive) found that merely providing AI tutoring tools is not enough to create meaningful engagement: in two districts studied, students' average weekly use of the AI tutor was just 2.18 and 5.23 minutes respectively - far below the 30 minutes per week the platform recommends for measurable reading gains.

Industry-pattern observations: generative-AI tutoring systems can deliver tailored practice, explanations, and follow-up questions, but reporting and research emphasize these systems do not address behavioral and motivational drivers of sustained practice. Tools that optimize problem difficulty, feedback timing, and personalization still frequently depend on external engagement mechanisms - classroom routines, teacher facilitation, parental support - to produce learning gains. The Stanford SCALE Initiative study found that even combining an AI tutor with a human tutor increased weekly engagement by only 1 to 4.4 minutes per session.

Reporting frames the slow uptake as a broader mismatch between technical capability and classroom dynamics. Limited engagement appears structural rather than technical: only about 53-61% of students even opened the AI tutor during scheduled time, per the Stanford study. Benefits appear more pronounced in some low-resource deployments, but evidence remains limited. Khan Academy has since overhauled its product to auto-activate Khanmigo during practice sessions rather than requiring students to seek it out.

Track rigorous pre-registered evaluations of AI tutoring deployed with different supports (teacher coaching, incentive structures, classroom integration). Monitor whether future research moves beyond access metrics to measure sustained usage, mastery, and transfer.