The AI-tutoring advantage holds across prior-ability and scientific-attitude subpopulations¶

Claim · All claims
Evidence · 1 study · causal r2 · q3
Currency · newest study 2025 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q3 i? Students both below and above 40% on pre-instruction FCI scores showed significantly better post-test performance with AI tutoring (p < 0.001), as did students below and above the 65% CLASS benchmark. → Kestin 2025

Evidence¶

Kestin 2025¶

Kestin, G., Miller, K., Klales, A., Milbourne, T. & Ponti, G. (2025). AI tutoring outperforms in-class active learning: an RCT introducing a novel research-based design in an authentic educational setting. Scientific Reports 15:17458. https://doi.org/10.1038/s41598-025-97652-6

q3 · i? · causal · r2

Subgroup analysis in the crossover RCT: significant subpopulations spanned the typical 30%-50% FCI range, and the improvement was similarly observed across the 65% CLASS scientific-attitude benchmark.

"The subpopulations of students with pre-instruction FCI scores below 40% and those above 40% both showed significantly better post-test performance with AI tutoring compared to in-class active learning (p < 0.001)."

Discussion¶

Related Claims¶

• Instruction type predicts post-test performance with a large effect size after controls (linear and quantile regression) — related
• Students learn more with a research-based AI tutor than with in-class active learning on identical content (post-test scores) — related
• Pre-instruction LCTSR scores correlate with normalized learning gains, most strongly for theoretical content (TUG-K r=0.59) and more weakly for descriptive content (DIRECT r=0.50) — reports the opposite