AI tools were supposed to make students think less, but a major university trial found a different result

AI tools were supposed to make students think less, but a major university trial found a different result

Summary

A preregistered randomized trial across 19 universities found that students in applied STEM classes given instructor-designed AI tools scored higher on proctored exams than students in classes awaiting rollout. The estimated gain was 0.32 standard deviations (95% confidence interval: 0.06 to 0.59). The trial included 7,200 students in 55 classes over three semesters; exam analysis covered 4,768 students across 44 course offerings. Researchers found no statistically significant change in engagement or satisfaction. The study did not compare the tools with generic chatbots, and questions remain about which designs drove the gain and whether it applies beyond applied STEM.

Latest News!