Inteligencia artificial en el arte rupestre: límites, posibilidades y desafíos de una evolución incierta

The recent development of artificial intelligence (AI) has opened new possibilities for the documentation and analysis of rock art. This article critically examines the potential and the limits of using open machine learning (ML) models for the automatic detection of pictorial motifs. After reviewing the state of the art and outlining the principles of ML as applied to archaeology, we empirically evaluate an open model (Roboflow 3.0 Object Detection) on four rock art panels from the Iberian Peninsula. The results show acceptable performance only in well-preserved contexts, while accuracy decreases notably in the presence of faded pigments or irregular rock surfaces. We conclude that ML can support archaeological research but cannot replace expert interpretation, and that progress in this field depends on standardized public repositories and coherent expert annotations.