Overview
- Stanford’s update on November 10, 2026 reviewed 52 randomized and quasi-experimental studies that test generative AI’s effects on students and teachers.
- Only 10 of the 52 studies involved U.S. K‑12 students or teachers, so the findings do not reliably generalize to most American schools.
- When students use AI to get explanations or practice problems they tend to learn more than when they use AI to produce finished work, but results across studies are mixed.
- Teacher-facing tools often increase feedback quantity and speed and can save time, yet at least one study found such tools lowered student motivation and confidence.
- Researchers and education coalitions are urging federally funded, carefully monitored trials and cautious district policies to build reliable, generalizable evidence before wide classroom rollout.