Corporate finance is a subject you learn by doing problems, and online students often do them alone, late at night, with no one to ask when they get stuck. Office hours only stretch so far.
The clever bit
Researchers at Boston University, Shihao Yang, Marshall Van Alstyne and Chrysanthos Dellarocas, gave students in an online MBA finance course access to an AI tutor that made them work through finance problems themselves. Students were randomly offered the tutor or not, so the two groups could be compared fairly. The study ran over the course's field period, with the last tests closing on August 15.
On the instructor's own final exam, students offered the tutor scored 2.6 points higher out of 100. That result covers all 86 students in the study, because every one of them took the final. The researchers' separate tests had far fewer completers, 51, so the final is the clearer measure.
Hooray
A few extra points on a final from practice that pushes students to think rather than copy is an encouraging sign that a patient, always-available practice partner can help adult learners who study on their own schedule. Because students were randomly offered the tutor, the gap between the groups is more likely down to the tutor than to which students chose to use it.
What comes next
It was one small course, students took part voluntarily, and the paper is a September preprint that hasn't been peer-reviewed. The study was preregistered, but the final-exam outcome wasn't one of the preregistered measures. A 2.6-point gain is modest, and it would need to show up again in other courses and other schools before anyone treats it as settled.