CoRemix: Supporting Informal Learning in Scratch Community With Visual Graph and Generative AI
Project Overview
The document highlights the implementation of CoRemix, an AI-driven system aimed at facilitating informal programming education within online communities, particularly Scratch. It targets the difficulties encountered by novice programmers, such as grasping intricate projects and applying fundamental computing principles. By leveraging generative AI, CoRemix offers personalized support and utilizes visual graphs to assist learners in breaking down projects into manageable components. A user study revealed that CoRemix not only significantly boosts learners' comprehension of computing concepts but also enriches their remixing practices, thereby indicating its effectiveness as a tool for enhancing informal learning in programming. Overall, the findings suggest that generative AI can play a vital role in improving educational outcomes in coding by providing tailored assistance and fostering a deeper understanding of essential concepts.
Key Applications
CoRemix
Context: Informal learning in online programming communities, targeting novice young programmers aged 9-12.
Implementation: CoRemix integrates visual graphs with generative AI, providing scaffolding and guidance for understanding programming projects and remixing practices.
Outcomes: Improved understanding of computing concepts, enhanced engagement in remixing activities, and better project understanding as shown by increased identification of key events and relationships.
Challenges: Initial cognitive load and potential errors in AI-generated content could confuse learners.
Implementation Barriers
Cognitive load
The integration of visual-textual scaffolding can increase learners' cognitive load, potentially hindering their ability to engage effectively.
Proposed Solutions: Balance between hands-on practice and guided support, ensuring learners have sufficient freedom to explore while receiving necessary scaffolding.
AI limitations
Generative AI may produce inaccurate or irrelevant responses (hallucinations) that can mislead learners. This can occur due to the limitations of the AI's knowledge base.
Proposed Solutions: Expand the knowledge base and apply the latest LLMs to improve response accuracy and relevance.
Project Team
Yunnong Chen
Researcher
Yishu Shen
Researcher
Ruiyi Liu
Researcher
Xinyu Yu
Researcher
Lingyun Sun
Researcher
Liuqing Chen
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Yunnong Chen, Yishu Shen, Ruiyi Liu, Xinyu Yu, Lingyun Sun, Liuqing Chen
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