Enhancing Coding Proficiency with Code Tailor

School of Information & College of Engineering

Problem

Assistant Professor Barbara Ericson aimed to improve the learning experience for students tackling close-ended write-code problems. Traditional methods often struggle to provide personalized, real-time feedback that aligns with students' individual solution paths. The challenge was to create a system leveraging U-M GPT Toolkit that could adaptively guide students through their coding errors and enhance their problem-solving skills

Audience

This U-M GPT Toolkit system was created for students enrolled in coding courses, particularly those in large classes where individualized feedback is challenging to provide.

Outcome/Impact

The implementation of Code Tailor significantly enhanced the coding proficiency and problem-solving skills of students. The system took students' incorrect code as input and generated corrected solutions that aligned with their original solution paths. If the generated solutions did not meet automatic evaluation criteria, Code Tailor iteratively refined the prompts to U-M GPT, ensuring high-quality code feedback. After three unsuccessful attempts, the system provided a low-level personalized solution to create effective mixed-up code puzzles.

This approach allowed students to receive immediate, tailored assistance, helping them understand and correct their mistakes in real-time. As a result, students developed a deeper understanding of coding concepts and improved their problem-solving abilities.