Abstract: Massive Open Online Courses (MOOCs) may reach a massive number of people, but few MOOCs count for credit. Scaling rigorous assessment, feedback, and integrity checks presents difficulties. We implemented an AI system for a CS1 MOOC-for-credit to address both scale and endorsement. In this analysis, we present the design of the system and an evaluation of the course. We observe that students in the online course achieve comparable learning outcomes, report a more positive student experience, and identify AI-equipped programming problems as the primary contributor to their experiences.
The full paper “Intelligent Evaluation and Feedback in Support of a Credit-Bearing MOOC” can be found here.