An Android-Based Multimodal AI Application for Contextual Environmental Learning in Children

Abstract

Children’s limited engagement with nature in the digital era poses a growing challenge for environmental education. This study presents the development of an Android-based educational application that leverages multimodal artificial intelligence (AI)—specifically the Google Gemini model—to facilitate contextual environmental learning for preschool and elementary-aged children. Using a prototyping methodology, the application integrates image capture, cloud-based processing through a FastAPI backend, and a Flutter-based interface designed for young learners. The system allows children to photograph plants and receive real-time, age-appropriate explanations about plant names, characteristics, and ecological functions in a narrative format. A limited usability trial involving children of varying age groups demonstrated positive engagement and curiosity, indicating the app’s potential as an interactive and enjoyable learning medium. Despite occasional inaccuracies in AI-generated descriptions and reliance on internet connectivity, user feedback suggested strong interest and educational value. Future enhancements will focus on developing localized plant databases, improving accuracy, and incorporating gamification elements. Overall, this study contributes to the growing field of AI-driven educational technology, demonstrating how multimodal AI can effectively bridge digital learning with real-world environmental experiences.