CodeTailor: LLM-Powered Personalized Parsons Puzzles for Engaging Support While Learning Programming
Project Overview
The document highlights the innovative use of CodeTailor, a generative AI tool aimed at improving education for novice programmers. By creating personalized Parsons puzzles, CodeTailor actively engages learners, prompting them to tackle coding problems rather than passively accepting solutions. Utilizing large language models (LLMs), the tool adapts to students' specific mistakes, providing tailored support that enhances both engagement and learning outcomes. Technical evaluations confirm that CodeTailor generates high-quality puzzles, while user studies reveal a clear preference among students for this interactive approach compared to straightforward AI-generated answers. Overall, the findings suggest that generative AI, exemplified by CodeTailor, can significantly enrich the educational experience by fostering active learning and improving student satisfaction and performance in programming.
Key Applications
CodeTailor: LLM-Powered Personalized Parsons Puzzles
Context: Introductory programming courses for novice programmers
Implementation: Students interact with the CodeTailor system by clicking the 'Help' button to receive personalized Parsons puzzles based on their incorrect code.
Outcomes: Increased engagement, improved application of programming concepts in posttest performance, and higher preference for using CodeTailor over AI-generated solutions.
Challenges: Some students struggle with understanding complex code blocks and occasional generation of overly complex solutions.
Implementation Barriers
Cognitive Barrier
Students may find it difficult to understand the details within individual blocks of code, hindering effective learning. CodeTailor occasionally generates complicated solutions that exceed learners' current knowledge.
Proposed Solutions: Adding explanations to each completed block or between blocks to enhance understanding. Automatic detection of overly complex code and adjustment of difficulty levels based on the student's progress.
Time Barrier
Students may experience delays in receiving real-time help from CodeTailor due to processing times.
Proposed Solutions: Optimize the waiting processes to improve the learning experience.
Project Team
Xinying Hou
Researcher
Zihan Wu
Researcher
Xu Wang
Researcher
Barbara J. Ericson
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Xinying Hou, Zihan Wu, Xu Wang, Barbara J. Ericson
Source Publication: View Original PaperLink opens in a new window
Project Contact: Dr. Jianhua Yang
LLM Model Version: gpt-4o-mini-2024-07-18
Analysis Provider: Openai