AI-accelerated Nazca survey nearly doubles the number of known figurative geoglyphs

Masato Sakai, Akihisa Sakurai, Siyuan Lu, J. Francisco Sibón Olano, Conrad M Albrecht, Hendrik F. Hamann, Marcus Freitag

Proceedings of the National Academy of Sciences 2024 article cited by 18

importance 4/10 solved Archaeology and Cultural Heritage archaeology stage: deployed

Finding figurative Nazca geoglyphs was extremely slow: only 430 were discovered over nearly a century, and small relief-type ones are hard to see.

Geoglyphs reveal ancient Nazca culture, and manual survey of a UNESCO World Heritage site is slow and error-prone.

An AI system was deployed across the entire Nazca region to detect candidate geoglyphs, which were then confirmed by field survey, specialising in hard-to-discern relief-type geoglyphs.

303 new figurative geoglyphs discovered in 6 months of field survey, nearly doubling the known total; relief-type motifs 81.6% human-related, average 43 m from ancient trails.

The AI-guided survey nearly doubled known figurative geoglyphs and enabled new analysis of their motifs and distribution.

Venue Proceedings of the National Academy of Sciences
Published 2024
Work type article
Authors Masato Sakai, Akihisa Sakurai, Siyuan Lu, J. Francisco Sibón Olano, Conrad M Albrecht, Hendrik F. Hamann, Marcus Freitag
Citations 18 (OpenAlex, at import time)
References 31
Model or method not stated in source
Organization not stated in source
Primary source https://doi.org/10.1073/pnas.2407652121
Open access hybrid
Keywords Nazca geoglyphs, AI survey, remote detection, archaeology, UNESCO, relief-type figures
4 / 10

real but narrower advance

Nearly doubled known figurative geoglyphs at a UNESCO site; major advance for Nazca archaeology but a narrow heritage domain, not broad societal scale.

Basis: stated. The source itself states the scale. Stated scale: 430 geoglyphs found over nearly a century; 303 new in 6 months. Method and limits.

# Paper Year Citations Importance
1 Machine Learning as a Tool for Hypothesis Generation
Jens Ludwig, Sendhil Mullainathan
2024 85
6
2 A Hybrid Semi-Automated Workflow for Systematic and Literature Review Processes with Large Language Model Analysis
Anjia Ye, Ananda Maiti, Matthew W. Schmidt, SJ Pedersen
2024 28
6
3 Restoring and attributing ancient texts using deep neural networks
Yannis Assael, Thea Sommerschield, Brendan Shillingford, Mahyar Bordba...
2022 206
5
4 Prediction on Domestic Violence in Bangladesh during the COVID-19 Outbreak Using Machine Learning Methods
Md. Murad Hossain, Md. Asadullah, Abidur Rahaman, Md Sipon Miah, Md. Z...
2021 41
5
5 Can Large Language Models Transform Computational Social Science?
Caleb Ziems, William Held, Omar Shaikh, Jiaao Chen, Zhehao Zhang, Diyi...
2023 559
4