A rigorous randomized controlled trial by Stanford's Graduate School of Education has found that low-income American students who used AI tutoring systems for 30 minutes per day gained the equivalent of two grade levels in math and 1.5 grade levels in reading over six months β results that exceed those of every other educational intervention ever studied at scale.
The study tracked 12,000 students across 80 public schools in California, Texas, and Georgia. Half used an AI tutor; half received standard instruction. The AI tutor group outperformed the control group by a margin that would typically require 18 months of in-class instruction.
The AI systems adapt in real time to each student's learning gaps, spending more time on concepts the student struggles with and less on material they have mastered β a form of individualized instruction that is impossible for a single teacher managing 30 students.
The Department of Education has announced a $2 billion initiative to deploy AI tutoring tools in Title I schools across all 50 states, reaching approximately 10 million low-income students by next school year.
