An Instagram-like XAI education tool effectively teaches K-12 students about AI-driven social media mechanisms through hands-on experimentation and real-time visualization.
Objective: To develop and evaluate an educational tool that teaches K-12 students about AI and data-driven mechanisms behind social media platforms, focusing on data collection, profiling, engagement metrics, and recommendation algorithms.
Methods:
- Created "Somekone," an Instagram-like interface with real-time analytics
- Tested with 209 children across 12 two-hour sessions
- Used paired devices: one for browsing, one for viewing analytics
- Implemented clustering analysis to study user behavior patterns
Key Findings:
- Identified three distinct user types: Browsers, Engagement Enthusiasts, and Selective Engagers
- Tool successfully demonstrated core AI concepts through hands-on experience
- Students gained understanding of data collection, profiling, and recommendation systems
- Real-time visualization helped students grasp complex AI concepts
Implications:
- Provides practical solution for teaching AI literacy to young students
- Bridges gap between social media use and understanding of underlying mechanisms
- Offers valuable tool for research on user behavior and algorithmic influence
Limitations:
- Simplified representation of complex social media mechanisms
- Maximum 30 concurrent users recommended
- Risk of students generalizing the tool's simplification as true representation
- Limited to Instagram-like interface
Future Directions:
- Evaluate learning outcomes at different levels
- Explore stronger integration of AI ethics
- Develop teacher materials and lesson plans
- Research user behaviors and engagement patterns
Title and Authors: "An XAI Social Media Platform for Teaching K-12 Students AI-Driven Profiling, Clustering, and Engagement-Based Recommending" by Nicolas Pope, Juho Kahila, Henriikka Vartiainen, Mohammed Saqr, Sonsoles López-Pernas, Teemu Roos, Jari Laru, Matti Tedre
Published On: December 18, 2024
Published By: arXiv