#randomized controlled trial

randomized controlled trial

3 papers

Turkish high-school students solve mathematics problems in three classroom conditions while teachers compare assisted practice with a later unaided test

Journal Paper2025

Unguarded GPT raised assisted practice scores but reduced unaided mathematics performance in a high-school RCT

Hamsa Bastani, Osbert Bastani, Alp Sungu, Haosen Ge, Özge Kabakcı, Rei Mariman

Proceedings of the National Academy of Sciences

In a preregistered cluster RCT with nearly 1,000 Turkish high-school students, a standard GPT-4 interface raised assisted practice performance by 48% but reduced later unaided exam performance by 17%. A teacher-informed tutor raised practice performance by 127% and largely removed the penalty without creating a positive exam gain.

generative AIhigh school mathematicsrandomized controlled trial

Read 500-word summary →

$Editorial cover of a middle-school learner using targeted mathematical feedback while a teacher reviews the learning process$

Conference Paper2026

Conference Paper 34

A Large Scale Randomized Control Trial Showing LLM Generated Feedback Helps Low-Knowledge Middle School Math Students with Short-Term Learning

Eamon Worden, Luca Dang, Wen-Chiang Ivan Lim, Sarah Miller, Jiayi Zhang, Aaron Haim, Adam Sales, Ashish Gurung, Neil Heffernan

ACM Learning @ Scale

Two large randomized trials in ASSISTments found that concise, misconception-targeted LLM feedback improved immediate correction and short-term transfer over correctness-only feedback, with the transfer benefit concentrated among lower-knowledge seventh-grade mathematics students.

LLM-generated feedbackmiddle school mathematicsrandomized controlled trial

Read 500-word summary →

Editorial cover for a secondary-school randomized trial of motivational and cognitive generative-AI learning prompts

Journal Paper2026

Journal Paper 32

Enhancing School Students' Self-Regulated Learning through Generative AI Support: A Randomized Controlled Trial

Tim Fütterer, Lisa Bardach, Jochen Kuhn, Stefan Daniel Keller, Peter Gerjets

Educational Psychology Review

A preregistered three-arm trial with 371 German students in Grades 7–9 found limited advantages for theory-informed GPT-4o prompts over standard ChatGPT, highlighting that meaningful engagement and implementation quality matter more than adding prompts alone.

self-regulated learningsecondary educationrandomized controlled trial

Read 500-word summary →