AI Archaeological Research: 10 Advances (2025)

AI is beginning to play a crucial role in enhancing archaeological research by offering innovative tools that streamline data analysis and uncover new insights.

1. Site Detection and Mapping

AI is increasingly used to scan vast areas via satellite imagery and LiDAR to pinpoint potential archaeological sites. By detecting subtle patterns or anomalies in the landscape that human eyes might miss, machine learning models can greatly speed up the discovery of sites. This allows archaeologists to focus on the most promising locations, saving time and resources. The approach is collaborative: AI quickly flags possible sites, and then experts verify them on the ground, integrating human judgment with computational power in site identification.

AI algorithms analyze satellite imagery and aerial photographs to identify potential archaeological sites, helping researchers locate dig sites more efficiently.

Site Detection and Mapping
Site Detection and Mapping: A researcher examining a large digital map on a computer screen, where AI highlights potential archaeological sites based on patterns detected in satellite imagery.

In a 2025 study, researchers trained a deep learning model (DeepLabV3+) on satellite images to recognize ancient Angkorian reservoirs in Cambodia. The AI achieved an F1-score of up to 45% in ideal conditions – comparable to a trained archaeology student – and could reduce manual mapping work by as much as 90%. By automatically highlighting likely archaeological features in remote sensing data, the system enabled the discovery of previously unknown water structures from the Khmer Empire, demonstrating how AI can vastly accelerate site detection.

Landauer, J., Klassen, S., Wijker, A. P., van der Kroon, J., Jaszkowski, A., & Verschoof-van der Vaart, W. B. (2025). Beyond the Greater Angkor Region: Automatic large-scale mapping of Angkorian-period reservoirs in satellite imagery using deep learning. PLOS ONE, 20(3), e0320452.

AI analyzes satellite images and aerial photography using advanced algorithms to detect anomalies in the landscape that may indicate the presence of archaeological sites. By processing large datasets rapidly, AI identifies patterns that are often imperceptible to the human eye, enabling archaeologists to pinpoint potential dig sites and plan expeditions more efficiently, thereby saving time and resources.

2. Artifact Analysis

Artificial intelligence aids in the analysis and classification of artifacts by using computer vision to recognize shapes, materials, and decorative patterns. Instead of sorting artifacts entirely by hand, archaeologists can leverage AI to quickly group pottery shards, lithics, or other finds into types or styles. This not only speeds up the cataloguing process for large collections but also increases consistency in identification. By training on thousands of known examples, AI systems can learn to spot minute differences or similarities that might elude a human, assisting researchers in interpreting artifact assemblages more efficiently.

AI aids in the classification and analysis of artifacts, using image recognition to quickly sort, identify, and catalog items based on shape, size, and patterns.

Artifact Analysis
Artifact Analysis: A detailed view of a computer screen displaying an AI system automatically categorizing various artifacts by shapes and materials, with digital tags and annotations.

A 2024 study applied deep learning to microscopic images of ancient ceramics to determine their origin. The ensemble AI model achieved roughly 96% precision and recall in classifying pottery pieces by their production source. In practical terms, the AI could correctly identify the kiln or region where a pot was made with about 95–96% accuracy, demonstrating near-expert performance in artifact classification. Such results show that AI can reliably analyze subtle texture and composition features of artifacts, augmenting experts’ ability to sort and study items from archaeological collections.

Wang, Q., Xiao, X., & Liu, Z. (2024). Using microscopic imaging and ensemble deep learning to classify the provenance of archaeological ceramics. Scientific Reports, 14, Article 32024.

Using image recognition technologies, AI assists archaeologists in the classification and analysis of artifacts. It quickly sorts through large quantities of items, identifying and cataloging them based on distinctive features such as shape, size, material, and decoration patterns. This automation speeds up the initial sorting process and ensures a high level of accuracy in artifact classification.

3. Predictive Modeling of Site Locations

AI-driven predictive models analyze environmental and historical data to forecast where undiscovered archaeological sites might lie. By learning patterns from known site locations – such as proximity to water, soil type, or elevation – machine learning algorithms can generate “heat maps” of high-probability areas for finds. This approach helps archaeologists target surveys in vast landscapes more effectively, focusing on spots that fit the profile of past human settlements. Predictive modeling doesn’t replace expert knowledge but rather enhances it, using big data to suggest likely site locations that archaeologists can then investigate in the field.

AI uses data from previous excavations and geographical information to predict where undiscovered archaeological sites might be located.

Predictive Modeling of Site Locations
Predictive Modeling of Site Locations: A scientist looking at a digital interface showing a predictive heat map of potential archaeological sites overlaid on geographic features, generated by AI analysis.

Researchers in 2023 developed a machine learning model to predict archaeological site locations in Japan and China by incorporating factors like terrain and water sources. The best model (a hybrid neural network combined with statistical filters) achieved an Area Under Curve (AUC) of around 0.90 in testing, indicating high accuracy in distinguishing areas with sites from those without. In practice, this AI model successfully highlighted that ancient sites cluster on certain landscapes – for example, in both regions it confirmed that sites tend to lie near water and on moderate slopes rather than flat plains. Such predictive mapping studies show AI’s power to learn non-random patterns in site distribution, guiding archaeologists to promising locations.

Wang, Y., Shi, X., & Oguchi, T. (2023). Archaeological predictive modeling using machine learning and statistical methods for Japan and China. ISPRS International Journal of Geo-Information, 12(6), 238.

AI leverages data from previously discovered sites and integrates it with environmental and geographic information to model and predict the locations of yet-to-be-discovered archaeological sites. This predictive capability allows researchers to focus their efforts more strategically and increases the likelihood of successful discoveries.

4. Restoration of Artifacts

AI is being used to virtually “restore” damaged artifacts and monuments by reconstructing missing pieces or original appearances. Using techniques like generative adversarial networks (GANs), AI can analyze what remains of an object – a faded painting, a broken sculpture, or an eroded artifact – and predict the intact version. This gives historians and the public a glimpse of how artifacts looked in their prime without physically altering them. Such reconstructions must be carefully validated by experts, but they offer valuable visual hypotheses for incomplete artifacts. Overall, AI-powered restoration provides a non-invasive means to aid conservation, education, and research by digitally reversing the effects of time.

AI algorithms can predict the original appearance of damaged artifacts or structures, providing visual reconstructions that help in understanding historical contexts.

Restoration of Artifacts
Restoration of Artifacts: A computer screen displaying a 3D model of a broken artifact next to its AI-predicted restoration, showing how the artifact would have originally appeared.

In 2024, scientists applied a generative AI model to enhance and reconstruct images of ancient Roman coins that had suffered heavy wear. The AI “learned” from thousands of well-preserved coins and then restored worn coin images so well that in some cases the results were virtually indistinguishable from originals. The system effectively re-rendered faded inscriptions and portraits on 1,700-year-old coin faces, making them look intact. This experiment demonstrated how AI can fill in missing details of artifacts: for these coins, the algorithm improved legibility and detail, aiding numismatists in reading and identifying features that manual restoration could not easily recover.

Altaweel, M., Khelifi, A., & Zafar, M. H. (2024). Using generative AI for reconstructing cultural artifacts: Examples using Roman coins. Journal of Computer Applications in Archaeology, 7(1), 301–315.

AI can reconstruct how damaged artifacts and structures might have originally appeared by analyzing their remnants and comparing them with similar intact findings. These AI-driven reconstructions provide valuable visual insights that help archaeologists and historians understand the original context and usage of these artifacts.

5. Deciphering Ancient Texts

AI is proving valuable in the decipherment and translation of ancient writings, especially for scripts and languages that are poorly understood or extinct. Machine learning algorithms can be trained on large corpora of known texts to recognize characters and suggest plausible translations for unknown texts. They excel at detecting patterns in symbols or letters – even if damaged or incomplete – and can match them to known languages or phrases. AI does in seconds what might take human epigraphers weeks, by scanning through dictionaries and prior translations. While human linguists are still crucial to interpret context and meaning, AI tools greatly accelerate the initial reading of ancient inscriptions or tablets, effectively acting as an automated translator for long-lost languages.

AI helps in decoding and translating ancient scripts by recognizing character patterns and suggesting possible meanings based on linguistic databases.

Deciphering Ancient Texts
Deciphering Ancient Texts: An archaeologist using a tablet to scan an ancient script, with the AI providing real-time suggestions for translation and interpretation on the screen.

A team from Israel trained a neural translation model in 2023 to convert cuneiform tablets written in Akkadian (a 5,000-year-old language) directly into English. The system was evaluated using a standard metric (BLEU) and scored ≈36.5, whereas human translators typically score around 40–50 on similar tasks. Notably, the AI achieved 97% accuracy at transliterating cuneiform signs into the Latin alphabet (essentially reading the script) before translating. This means the program can render the wedge-shaped cuneiform signs into legible text almost perfectly, and then produce a rough English translation nearly on par with expert Assyriologists. The tool has already been used to instantly translate tens of thousands of digitized clay tablets, a process that previously required scarce specialists and considerable time.

Gutherz, G., Gordin, S., Sáenz, L., Levy, O., & Berant, J. (2023). Translating Akkadian to English with neural machine translation. PNAS Nexus, 2(5), pgad096.

AI enhances the decipherment of ancient texts by using pattern recognition to identify recurring symbols and characters. It correlates these with known languages and scripts to suggest possible translations. This tool is particularly useful in cases where the language has not been fully decoded, offering new ways to unlock historical narratives.

6. 3D Modeling and Simulations

AI technologies assist in creating detailed 3D models of archaeological sites and artifacts, which can be used for virtual reconstructions and simulations. By feeding algorithms with photographs or laser scans, archaeologists can generate digital twins of ruins or objects. These 3D models allow researchers to explore sites as they were in the past, simulate environmental effects, or even digitally test how structures might have looked or functioned. AI improves this process by filling gaps in data; for example, it can infer a 3D shape from partial remains or even from old 2D images. The resulting virtual models are invaluable for public education (through virtual tours) and for scholars to analyze spatial relationships at a site without being physically present.

AI assists in creating 3D models of archaeological sites and artifacts, which can be used for virtual tours, detailed analysis, and educational purposes.

3D Modeling and Simulations
3D Modeling and Simulations: An interactive 3D model of an archaeological site displayed on a digital workstation, with a researcher using AI tools to simulate different historical scenarios.

In 2024, researchers demonstrated an AI method that can produce a 3D reconstruction of an artifact from a single old photograph. Using a 134-year-old black-and-white photo of a now-hidden temple relief at Borobudur (Indonesia), their neural network generated a three-dimensional digital model of the relief. The AI effectively “lifted” the scene out of the photograph, recreating depth for features like carved figures and backgrounds that had long been covered by restoration work. This proof-of-concept shows how AI-driven 3D modeling can recover the form of artifacts or structures that no longer exist in view, enabling archaeologists to examine and even virtually restore heritage that is inaccessible or lost.

Pan, J., Tanaka, S., et al. (2024). AI-based single-image 3D reconstruction of a hidden Borobudur relief (Conference presentation). 32nd ACM International Conference on Multimedia.

AI facilitates the creation of detailed 3D models of archaeological sites and artifacts. These models can be used for more than just virtual tours; they provide a platform for conducting simulations, engaging in digital preservation, and sharing detailed replicas with the global academic community, enhancing collaborative research and education.

7. Temporal Analysis

AI tools help archaeologists analyze the chronological sequence of artifacts and site layers, refining our understanding of timelines. By examining patterns in stratigraphy (the layers of soil deposits) or in artifact styles over time, machine learning can assist in dating finds and correlating layers across different areas of a site. For example, AI can detect subtle changes in pottery style progression or in sediment composition that indicate different time periods. Additionally, AI algorithms can integrate large datasets of radiocarbon dates, artifact typologies, and stratigraphic diagrams to model a site’s occupation history. In essence, AI acts as a sophisticated statistical assistant, finding order in complex temporal data and helping archaeologists build more precise chronologies of past human activity.

AI examines the stratification of archaeological sites to help determine the chronological sequence of historical layers, enhancing the understanding of how a site was used over time.

Temporal Analysis
Temporal Analysis: A digital timeline on a monitor showing the stratification of an archaeological site with different layers dated and analyzed by AI.

A 2024 study used deep neural networks to automatically estimate the age (archaeological period) of artifacts based on their images. The researchers trained the model on over 10,000 artifact photos spanning 16 defined periods from the Paleolithic to Late Islamic times. In tests, the AI’s top-5 accuracy (likelihood that the correct period was among its five highest guesses) exceeded 90%. In other words, the system could correctly place an artifact into its right general era (e.g. Bronze Age, Iron Age, etc.) with high reliability, approaching the success rate of veteran archaeologists. Such performance underscores how AI can assist in temporal classification; by recognizing diagnostic features (like tool shapes or decoration styles) it provides quick initial dating suggestions that experts can then refine.

Yalov-Handzel, S., Cohen, I., & Aperstein, Y. (2024). Comparative analysis of CNN architectures and loss functions on age estimation of archaeological artifacts. Journal of Computer Applications in Archaeology, 7(1), 185–194.

AI examines the layers of soil and deposition at archaeological sites to help determine the chronological order of cultural deposits. By automating the analysis of stratification, AI aids archaeologists in reconstructing the timeline of how a site was used, which is crucial for understanding the historical context of the findings.

8. Analysis of Human Remains

AI is being applied in bioarchaeology to extract more information from human skeletal remains. By training on medical and forensic datasets, AI can learn to determine characteristics like age-at-death, biological sex, stature, or even ancestry from bone measurements or scans. It can also assist in detecting signs of ancient diseases or dietary deficiencies by recognizing patterns in skeletal lesions or tooth wear. These algorithms augment the work of osteologists by providing quick, objective initial assessments – for instance, scanning a skull to predict sex – which can then be confirmed by experts. The result is a more efficient analysis of burials and human remains, leading to richer insights into past populations’ health, diet, and demographics.

AI techniques in bioarchaeology help determine age, sex, diet, and health of human remains found at archaeological sites, providing insights into past human populations.

Analysis of Human Remains
Analysis of Human Remains: A forensic anthropologist examining a digital screen that displays AI-assisted analysis of human skeletal remains, including age, diet, and health indicators.

A recent AI tool has achieved remarkable accuracy in assessing skeletal remains. In a 2024 study, a deep learning model was trained on 3D CT scans of skulls and correctly determined the biological sex in 97% of cases, compared to about 82% accuracy for experienced human osteologists. The model was not only more accurate but also about five times faster than a person at evaluating key cranial features. This high precision means that, for example, by analyzing a digitized skull from an archaeological dig, the AI can rapidly identify it as male or female with minimal error. Such systems, extended to other bones or traits, are poised to assist archaeologists in constructing profiles of ancient individuals (age, sex, stature) with greater speed and consistency.

Lye, R., Min, H., Dowling, J., Obertová, Z., Estai, M., Bachtiar, N. A., & Franklin, D. (2024). Deep learning versus human assessors: forensic sex estimation from three-dimensional CT scans. Scientific Reports, 14, Article 30136.

In bioarchaeology, AI techniques are applied to analyze human remains more precisely. AI can assess factors such as age at death, biological sex, diet, disease, and overall health based on measurements and patterns identified in skeletal remains, offering deeper insights into the lives of past populations.

9. Remote Sensing Data Analysis

Modern archaeology often relies on remote sensing technologies – such as aerial photography, LiDAR, radar, and satellite imagery – to detect buried structures and landscape modifications. AI greatly enhances the analysis of these massive datasets by automatically identifying patterns indicating archaeological features (e.g., geometric shapes of ruins, mounds, ancient field systems) that might be invisible at ground level. For example, AI can scan LiDAR’s 3D point clouds of a jungle and outline rectilinear structures or road networks hidden under vegetation. By doing so, AI turns what would be years of manual image inspection into a much faster, repeatable process. The outcome has been the discovery of “lost” cities and features around the world, as AI-guided remote sensing peels back layers of earth or forest to reveal the human imprints beneath.

AI processes data from remote sensing technologies, such as LIDAR, to uncover hidden features of landscapes, revealing buried structures or modifications made by ancient civilizations.

Remote Sensing Data Analysis
Remote Sensing Data Analysis: Aerial view of a landscape on a computer screen with AI-identified sub-surface features visible through LIDAR data, highlighting ancient structures hidden beneath vegetation.

One dramatic example comes from the Maya Lowlands: in 2022, researchers used AI-assisted LiDAR analysis to map a vast Preclassic (~1000 B.C.–A.D. 150) Maya civilization in northern Guatemala. The LiDAR survey identified over 770 ancient settlements, which archaeologists consolidated into 417 distinct cities, towns, and villages connected by a network of raised causeways totaling about 177 km (110 miles) in length. This hidden landscape, complete with pyramids, plazas, and roadways, had been completely obscured by dense rainforest. AI algorithms helped to filter and interpret the LiDAR data, pinpointing structural outlines beneath the canopy. The discovery redefined the scale of Maya urbanization in that region – a feat only possible by coupling remote sensing with intelligent data processing on an unprecedented scale.

Hansen, R. D., Schreiner, T., Suyuc Ley, E., Martínez, G., Morales-Aguilar, R., Richards-Rissetto, H., … & Estrada-Belli, F. (2022). LiDAR analyses in the contiguous Mirador-Calakmul Karst Basin, Guatemala: new perspectives on Preclassic Maya organization. Ancient Mesoamerica, 33(3), 1–16.

AI processes large volumes of data from remote sensing technologies, such as LIDAR, which penetrates forest canopies to reveal ground features. AI's ability to analyze this data efficiently helps uncover hidden structures and landscape modifications made by ancient civilizations, providing a clearer picture of historical land use.

10. Automated Documentation

AI is streamlining the tedious task of documentation during archaeological excavations. Traditionally, archaeologists record every find and feature by hand – noting its location, depth, description, etc. Now, intelligent systems are being introduced to help automate this process. For instance, tablet-based field apps with AI can auto-fill data entries when an artifact is photographed or scanned, or even transcribe spoken observations by archaeologists into structured notes. Some systems use object recognition to identify artifacts on the spot and log their details. By organizing excavation data in real-time and reducing human error, AI-driven documentation ensures more complete and accessible records. This means that the wealth of information from a dig (photos, maps, notes, artifacts) can be systematically captured and linked in a database, often ready for analysis immediately after – or even during – the excavation.

AI streamlines the documentation process by automatically recording data during excavations, organizing information into databases, and ensuring detailed and accurate record-keeping.

Automated Documentation
Automated Documentation: A field archaeologist using a rugged tablet to automatically input excavation data into a digital form, with AI organizing and storing information systematically in a cloud database.

A European project developed an AI-powered application called ArchAIDE to automate pottery identification and recording in the field. Using a single photo of a pottery fragment, the ArchAIDE app’s image-recognition AI can identify the type of ceramic and instantly populate a digital catalog with its attributes. In trials, this tool markedly reduced the time required for archaeologists to classify and document pottery finds, compared to purely manual cataloguing. For example, a task like identifying a Roman amphora shard – which might take an expert several minutes with reference books – can be done in seconds by the app, which then stores the identification, context, and even suggested bibliography in a database. By speeding up artifact logging and minimizing misclassification, such AI-assisted documentation tools help ensure that no data is lost in fast-paced excavations and that researchers can query the information immediately.

Anichini, F., Gesquière, G., Cornwall, K., Konoplicki, S., Tudisco, F., & Scopigno, R. (2019). Developing the ArchAIDE application: A digital workflow for pottery identification. Internet Archaeology, 52.

During excavations, AI helps in the automated documentation of findings. It systematically records details about the location, depth, and context of each artifact or feature uncovered, ensuring that all data is meticulously cataloged and easily accessible for future analysis. This streamlined approach improves the accuracy of field reports and enhances the archival process.