Intelligent Guide Systems for Digital Heritage: A Systematic Review of AR-AI Integration in the Era of Generative Cognition

Abstract

Digital transformation in cultural heritage institutions has increasingly prioritized immersive technologies to enhance visitor engagement and knowledge delivery [8]. However, traditional Augmented Reality (AR) solutions remain constrained by the scalability bottlenecks of marker-based tracking and the pedagogical limitations of static content delivery [1]. This work uses the Grand Egyptian Museum (GEM) as a primary case study to validate a consolidated system-level perspective of recent advancements in the integration of Artificial Intelligence (AI) with AR [6]. By synthesizing existing research through the lens of GEM’s large-scale operational challenges—such as an expected annual volume of 5–8 million visitors—this study identifies recurring architectural patterns. It evaluates how robust perception capabilities, achieved through mobile-optimized Convolutional Neural Networks such as YOLOv8 and EfficientNet [2, 15], and intelligent cognition, powered by Large Language Models (LLMs) grounded in historical Knowledge Graphs [4, 19], address critical technical challenges including lighting sensitivity and occlusion [9]. The findings indicate that the convergence of adaptive vision and generative storytelling transforms the museum visit from passive observation into an active, dialogue-driven exploration [25]. The study concludes that applying this framework to the Grand Egyptian Museum establishes a new benchmark for ’Generative Cognition’ in high-traffic heritage sites, providing a scalable model for transforming visitor interaction globally.