Generales

By Andrea Carolina Vargas Malagón, Journalist at UdeA Communications Office 

Artificial intelligence (AI) in archaeology integrates traditional techniques with modern technology to refine processes and accelerate analysis, giving researchers more time to rethink historical interpretations.  

Generative artificial intelligence learns patterns and relationships from a human-created dataset using a machine learning model / AI-generated image

For over 200 years, archaeologists have explored sites, uncovered artifacts, and reconstructed history to trace the evolution of human societies. As in other sciences, new tools and methods have refined data analysis, reduced costs, and freed researchers to deepen their theories on past eras. The integration of generative artificial intelligence was only a matter of time. 

“Artificial intelligence enhances archaeology by streamlining the analysis of vast datasets, reducing the time spent describing and categorizing artifacts, and allowing researchers to dedicate more time to interpretation,” said Daniel Grisales Betancur, a master’s in medieval archaeology and professor at the Faculty of Social and Human Sciences at the Universidad de Antioquia. 

“Artificial intelligence is a branch of computer science comprising various subfields, all focused on developing systems or programs that can perform tasks usually requiring human intelligence,” explained Daniel Escobar, a master’s in telecommunications engineering and professor at the UdeA School of Engineering. 

Archaeology is a social science that seeks to understand human behavior over time. However, researchers cannot simply collect data from archaeological remains; they must gather, organize, and analyze it to develop theories about the past. It is where technological advancements, including AI, from the exact sciences become essential. 

“Unlike many other social sciences, archaeology has always been strongly connected with the exact sciences. While people often romanticize the image of an archaeologist digging alone in the jungle, the truth is that archaeological knowledge develops through natural science application principles,” commented Daniel Grisales.  

Descriptive statistics, osteology, archaeometry, remote sensing, and radiocarbon dating are among the traditional techniques that preceded the use of artificial intelligence. Initially novelties in the field, archaeologists gradually accepted these tools over time as they proved their value. However, the convergence of the distant past with an AI-driven future may offer new paths to understanding the civilizations that came before.  

Remote sensing and radiocarbon dating  

Remote sensing is a technique that collects information about the Earth’s surface without physical contact, primarily using aerial and satellite images. Radiocarbon dating is a scientific method for determining the age of carbon-based materials, such as wood, bone, charcoal, textiles, and other organic substances, up to around 50,000 years.  

“Artificial intelligence offers itself as a tool to archaeology, one that will be embraced or rejected not because of AI’s limitations, but due to the limitations in archaeological thinking. The challenges we face with AI are the same as those with any technique from the natural sciences. Archaeologists often mistake the tool for the knowledge itself when artificial intelligence is just a means to an end,” said Grisales.  

Other findings with AI  

Several examples showcase the partnership between artificial intelligence and archaeology. This year, AI helped an archaeologist group uncover nearly as many geoglyphs in the Nazca desert of Peru in a few months as were found in the past century. Last year, AI facilitated the discovery of new sites across 475,000 km² of the Indus River Valley—covering Afghanistan, Pakistan, and northwest India—and identified about 6,000 archaeological sites, highlighting potential areas for historical research.  

Artificial intelligence offers immense potential for archaeology, from identifying patterns in satellite images that suggest buried structures to processing spatial data for creating predictive models of archaeological sites. However, it also presents significant challenges. These include developing algorithms capable of handling the vast diversity of archaeological data, ensuring the quality and integrity of the data used to train AI models, and ensuring that archaeologists acquire the necessary expertise to employ these advanced tools effectively. 

Studying the complexity of the past   

As part of the effort to integrate artificial intelligence into the study of the past, Daniel Sanchez, an anthropologist from the Universidad de Antioquia, a PhD candidate in Archaeology at the University of Lisbon, and a member of the Quantitative Archaeology Group at both the University of Lisbon and the University of Seville, led the creation of an AI model designed to identify the minerals in personal ornaments found in excavations, particularly in the Iberian Peninsula.