ClawAI: A Software Tool Teaching Text Classification and AI Systems to Middle School Students

Machine learning (ML) is a transformative technology shaping the modern world, yet its concepts remain inaccessible to many, especially to middle school students. To address this gap, we created an educational game, ClawAI, that introduces AI text classification through interactive play. ClawAI was built using python and connecting our game to the Machine Learning for Kids platform using a Watson API Key. The game engages children in describing objects for an AI claw machine to identify. Students learn key concepts like confidence levels and text classification across three rounds of gameplay. Tested with 43 students aged 7–14, the game demonstrated its effectiveness in teaching AI concepts, with 34 participants showing understanding of text classification through gameplay. This suggests our game successfully bridges the gap between AI education and engagement, offering a scalable model for teaching machine learning in K–12 settings. We had two groups of students, one group that were visuals, and the other group were conceptual students. These two strategies emerged from successful and unsuccessful students. In our findings we found two distinctive groups being formed based on the way they played and the words they used to classify these items. Overall half of the students played the game well and proved how they understood how text classification works with AI, while the other half did not. Out of the half that did well, two-thirds of them showcased a learning experience. Our results demonstrate that our game and games alike can be utilized as a learning tool when students are taught about AI in K-12 education.