2026 ASEE Annual Conference & Exposition

Programming projects have long been a cornerstone of core CS courses, engaging students in higher-order cognitive skills such as applying concepts, analyzing algorithmic approaches, and evaluating program outputs. The increased use of autograders is beneficial for providing rapid feedback and grading efficiency, but often necessitates fully-specified, highly-scaffolded project designs, which may constrain opportunities for open-ended problem solving and independent decision-making. These concerns are amplified by widespread availability of generative AI (genAI) tools, which can often produce solutions quickly. In order to investigate the impact of incorporating open-ended components into otherwise fully-specified programming projects in a second-year programming course, two counterbalanced programming projects, centered on word puzzle games, namely Spelling Bee and Word Ladder, are used to compare student experiences, learning gains, and independence. Data collection instruments focus on student perceptions, concept proficiency, self-efficacy, belongingness, and development/submission behavior.
Results show that students prefer fully-specified project tasks, open-ended project tasks are more stressful, but open-ended project tasks lead to a stronger sense of accomplishment. Participants self-report that they tended to utilize genAI tools for idea generation and debugging during project development. Student feedback on perceptions, AI usage, and project experience give more insight to the trends found within the data.