Artificial Intelligence for Cultural Heritage Characterization: Towards Scalable and Sustainable Territorial Inventories

The depopulation of rural areas across Southern Europe has accelerated the deterioration and disappearance of cultural heritage assets, many of which remain undocumented, undervalued, or at risk. This paper presents a hybrid methodology that combines artificial intelligence (AI) techniques with expert knowledge to improve the identification, classification, and validation of cultural heritage in rural territories. The proposed system processes unstructured textual data to extract and categorize information on tangible and intangible heritage elements. A pilot case study was carried out in the Valle de Amblés and Sierra de Ávila (province of Ávila, Spain), a region characterized by severe demographic decline but high heritage potential. Four municipalities (Amavida, Cardeñosa, La Torre, and Solosancho) were selected for testing based on heritage diversity, available documentation, and accessibility. In each case, AI-generated outputs were compared with manually curated validation files to assess the detection performance and applicability of the system. Results show that while traditional fieldwork and historical interpretation remain essential, AI can significantly enhance the speed and scalability of heritage documentation, particularly in under-resourced areas. The methodology reinforces the value of interdisciplinary collaboration and supports the development of territorial heritage strategies for sustainable rural regeneration.