PolluSpot: A Novel Tool for Teaching Model Training to Middle School Students

The increasing integration of artificial intelligence (AI) into everyday life necessitates early exposure of individuals at a young age to its foundational concepts. PolluSpot, an innovative educational tool, was developed to engage middle school students in learning supervised machine learning through decision tree models. Using real-world environmental datasets of polluted and unpolluted images, PolluSpot combines interactive learning with environmental awareness. Students train the AI by labeling images as either polluted or unpolluted based on features such as smoke, garbage, dirty water, dull, grass and tree, they observed how their input influences the model’s predictions. This study evaluates PolluSpot’s effectiveness in enhancing AI literacy among students in grades 6-8. Results indicate significant improvement in students’ understanding of AI concepts, with 74% of the students demonstrated ability to describe how AI learns, and 88% accurately understood classification processes.