Desarrollo de un sistema educativo basado en dinámicas gamificadas para la enseñanza de la cultura Cañari con generación automática de narrativas usando un modelo multimodal

Abstract

The project develops an innovative digital ecosystem to rescue knowledge about Cañari cultural heritage among Ecuador's younger generations. The solution integrates gamification with artificial intelligence through the multimodal LLaVA-1.5-7B model, fine-tuned with 1,000 steps to rigorously interpret the Cañari archaeological legacy. The architecture implements a hybrid SOA model with a Kotlin backend managing game logic and a Flask server processing AI inferences in the cloud, allowing users to explore sites like Ingapirca where the AI identifies cultural elements in real-time and generates instantaneous educational narratives. User validations through SUS and TAM yielded excellent results: SUS score of 77.32 (excellence in usability), Perceived Usefulness of 4.46/5.00, and Ease of Use of 4.50/5.00, confirming that this integration of multimodal AI and gamification surpasses traditional methods and effectively connects ancestral knowledge with the contemporary digital citizen.