In a randomized trial across 19 universities, students in applied STEM courses performed better on proctored exams after instructors introduced AI tools tailored to their classes. The trial enrolled 7,200 students across 55 classes, while the exam analysis included 4,768 students from 44 course offerings. The estimated effect was 0.32 standard deviations, with a 95% confidence interval of 0.06 to 0.59, equivalent to roughly 33% of the typical spread in scores within a class. The tools included tutors, coaches, interactive simulations and role-playing exercises, with no statistically significant change in engagement or satisfaction. The study nevertheless remains limited to applied STEM courses and does not directly measure performance against generic chatbots.
Source: Read the original article


