Generative AI as an Adaptive Narrative Layer for Cultural Heritage: Towards Responsible Personalized Storytelling

Recent advances in Generative Artificial Intelligence have introduced unprecedented possibilities for dynamically producing personalized content. In cultural heritage contexts, these technologies enable a shift from static interpretive materials toward adaptive, context-sensitive narrative experiences. Generative systems promise increased engagement and accessibility for museums, archaeological sites, historic environments, natural heritage locations, and digital archives. However, their use raises significant epistemic and ethical concerns. Cultural heritage interpretation is not merely informational, but deeply connected to identity, memory, and institutional authority. This paper proposes a vision of Generative AI as an adaptive narrative layer embedded within curated heritage systems. Rather than replacing curatorial expertise, generative systems should operate within bounded, knowledge-grounded frameworks that preserve authenticity and trust. We outline key research challenges related to user modeling, contextual adaptation, epistemic integrity, governance, and evaluation, and propose some research challenges and directions for responsible generative personalization in cultural heritage.