SchoolAI makes students think

SchoolAI's research team analyzed 13,882 student-AI conversations from 82 teachers in Utah's Jordan school district. The team compared October 2023, shortly after the platform launched, with October 2025, after two years of classroom use, rating every conversation on a four-level Bloom's Taxonomy scale that ranges from basic recall up to analysis, evaluation, and creation.

Critical thinking scores rose 28.3%. Level 1 conversations, the ones with no observable critical thinking, dropped from 76.7% to 58%, while Levels 3 and 4, the deepest analytical and evaluative thinking, more than doubled.

By 2025, nearly half of all student conversations showed higher-order thinking, and the pattern held across every subject and grade level studied.

Meaningful teacher engagement drove the gains. Educators who designed six or more unique learning spaces on the platform saw significant improvement in their students' thinking, while those with minimal use did not. Even the remaining Level 1 conversations looked different by 2025: choice-based activities, where students make decisions in branching scenarios or simulations, rose from 20.1% to 36.2%, a sign that lesson design itself was shifting toward intentional engagement.

For education leaders, the gains came from a specific combination: teachers designing learning experiences where AI guides the thinking instead of replacing it. Access to AI alone wasn't enough.

Download the full report, which received ESSA Tier 3 Certification through Instructure, for the complete methodology and findings.