Development and Utilization of AI-Enabled Automatic Programming Problem Generator Using the CodeRunner Plugin of Moodle
Jay Vince Donoso Serato, Cherry Lyn C. Sta. Romana · 2024
Programming instructors provide various kinds of problems that suit their current topics of programming. With the use of Learning Management Systems (LMS) such as Moodle, teachers can create and store their problems in problem banks using a question plugin CodeRunner. However creating and validating programming problems takes significant time and creative effort. With instructors dealing multiple programming languages, it would need mastery not only to solve the problem but also to use a specific language. This paper presents an automation of the programming problem generation using AI. This is effectively an improvement of the CodeRunner plugin of Moodle that allows programming instructors to generate problem descriptions, answers, and testcases of different programming topics in various programming languages with a few clicks. To address the issue of difficulty control, an additional feature is placed to generate an easier or harder problem than what is currently generated. The evaluation of the improved tool showed that the problems generated with AI are similar, correct, and practical that matches the human-generated problems.