Leveraging AI to Enhance Educational Practices
Role: Postdoctoral researcher in artificial intelligence at LIG (Laboratoire Informatique de Grenoble)
Collaborators :
- Mathieu Pinelli and Salomé Cojean – Researchers at LaRAC (Laboratoire de Recherche sur les Apprentissages en Contexte)
- Nassim Bouarour and Sihem Amer-Yahia– Researchers at LIG
Funding
This project is supported by the MIAI Grenoble Alpes Institute under the TUPAIA Chair (Technologies Used in Pedagogy with Artificial Intelligence Adaptations).
The TUPAIA Chair aims to integrate artificial intelligence into educational practices to:
- Enhance Personalized Learning: Adapt educational content to individual learner characteristics by considering cognitive variables such as motivation, cognitive load, and satisfaction.
- Improve Acceptability of AI-Powered Tools: Investigate factors influencing the acceptance of AI-driven educational tools among users, including teachers and students, to ensure effective implementation.
Research Projects
1. Impact of Note-Taking and Breaks on Cognitive Load and Performance
Main collaborator: Mathieu Pinelli
Objective: Examine how note-taking strategies and breaks affect learners’ cognitive load and performance during video-based learning sessions.
Methodology:
- Subgroup Discovery (SD): dentifying patterns to determine which factors influence cognitive load, learning performance, and learning strategies
- Gradual Pattern Analysis: Analysis of the covariances of variables with different types of cognitive load to identify those that significantly influence cognitive processing.
Results: This study provided deeper insights into learning strategies by identifying key variables affecting cognitive load. One key finding was that pauses mediate the effect of note-taking on learning performance – more notes led to increased pauses, which in turn improved learning outcomes.
2. Personalized Video Recommendations Using Reinforcement Learning
Main collaborator: Nassim Bouarour
Objective: Develop a system that provides personalized video content recommendations based on individual learner profiles, focusing on cognitive load and skill levels.
Methodology:
- Deep Q-Networks (DQN): Implement reinforcement learning techniques to tailor video recommendations that align with learners’ cognitive states and competencies.
Outcomes: Creation of adaptive learning environments that enhance engagement and knowledge retention.
Project Status: Data collection is ongoing to refine the model and assess its effectiveness in improving learning engagement.
Dissemination
Journal Publication
📄 Pinelli. M., Moranges. M. & Cojean. S. Take a break, take notes? Effect of pauses and note-taking on performance learning in instructional videos. Submitted Journal paper
