Generating Personalized Assignments with Students in the Loop

We present a system that enables students to generate their own programming assignments that are both personalized to their interests and aligned with the course objectives. A group of twelve undergraduate students randomly selected from an introductory programming course (~200) received training and access to the system for six weeks. Students were free to use the system to generate assignments or take regular assignments instead. Weekly student feedback was used to refine the system through three iterations, and a focus group with three students was held after the course to gather opinions on the experience. Our findings are complicated and cautionary: students appreciated generating their own personalized assignments with feedback on demand. However, the cognitive load of having to generate and decide which assignment to complete, along with the variation of difficulty and alignment with learning objectives were noted as problematic and discouraging. Perhaps the most negative but heartwarming result is that students missed the community aspect of independently solving and collectively discussing solutions to a common assignment.