AI Assistance for Discretionary Work: Increasing Feedback Provision in Higher Education

Abstract

AI assistance can increase the efficiency and accuracy of human work, but can it also encourage people to undertake valuable tasks they might otherwise skip? Here, we investigate whether AI assistance can improve participation in discretionary but beneficial work in the context of personalized feedback provision in higher education, a pedagogically valuable but often optional practice. We conduct a mixed-methods study combining a randomized field experiment and qualitative interviews in an upper-level machine learning course with $n$=11 teaching assistants (TAs) and n=88 students. Student submissions were randomly assigned to either (1) a treatment condition where TAs received AI-assisted feedback drafts after grading or (2) a control condition without drafts. We find that AI-assisted feedback significantly increases feedback provision (+10.8 percentage points, SE=1.1, p<0.001) and feedback length (+39.8 chars, SE=3.45, p<0.001) while also maintaining student-rated usefulness and expert-rated feedback quality among the feedback instances for which these outcomes were observed. While time spent conditional on providing feedback did not significantly differ between conditions, total time spent in the feedback stage increased in the treatment condition (10.55 seconds, SE=4.19, p=.012). Qualitative findings suggest that AI-assisted drafts function as editable scaffolds that lower barriers to initiating feedback rather than reducing overall labor. Together, our findings show that AI assistance can increase participation in discretionary communication without necessarily reducing the labor involved or eliminating the effort required to sustain it.