JELAI: Integrating AI and Learning Analytics in Jupyter Notebooks
Project Overview
The document explores JELAI, an innovative open-source platform that merges Learning Analytics with Large Language Model-based tutoring in Jupyter Notebooks, showcasing the transformative potential of generative AI in education. It emphasizes the importance of pedagogical grounding and context-aware AI interactions in maximizing the effectiveness of such technologies. JELAI excels in capturing detailed student activities and chat interactions, enabling adaptive, real-time interventions tailored to individual learning needs and fostering research into student behavior patterns. Through two proof-of-concept use cases, the platform illustrates its capabilities in logging data and analyzing help-seeking behaviors among students, which ultimately supports enhanced educational outcomes. The findings underscore generative AI's role in personalizing learning experiences and providing timely assistance, while also highlighting the challenges that need to be addressed for successful integration in educational settings.
Key Applications
JELAI (Jupyter Environment for Learning Analytics and AI)
Context: Educational context within Jupyter Notebooks, targeted at students learning programming, specifically in a Python course.
Implementation: Implemented as a modular, containerized system using open-source technologies, allowing integration of LA with LLM tutoring.
Outcomes: Successfully captured granular data streams, enabled real-time context-sensitive AI scaffolding, and facilitated A/B testing of AI configurations.
Challenges: Initial studies require validation with larger samples; computational cost of powerful local LLMs; technical expertise needed for configuration.
Implementation Barriers
Technical Barrier
The computational cost of powerful local LLMs and the technical expertise required for configuration pose challenges to implementation.
Proposed Solutions: Future work includes enhancing AI interaction, simplifying configuration, and improving interoperability.
Project Team
Manuel Valle Torre
Researcher
Thom van der Velden
Researcher
Marcus Specht
Researcher
Catharine Oertel
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Manuel Valle Torre, Thom van der Velden, Marcus Specht, Catharine Oertel
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