Archaeologists surveying the Nazca Pampa in southern Peru have identified 303 previously unknown geoglyphs with the help of artificial intelligence, a haul that nearly doubles the tally of known relief-style figures in one of the most heavily studied desert landscapes on Earth. The work, a collaboration between Japan’s Yamagata University and IBM Research, was published in the Proceedings of the National Academy of Sciences and, according to the reporting on the study, turned up the new figures in roughly six months of image analysis rather than the years such fieldwork usually demands.
Why a near-doubling of the Nazca figures matters
The Nazca Lines have drawn scientists and travelers for nearly a century, a sprawling gallery of birds, monkeys, geometric shapes and human-like forms etched into a desert plateau by scraping away dark surface stones to expose the lighter ground beneath. The largest are unmistakable from the air; many others are faint, small and easy to walk past. For decades researchers hunted the hidden ones with aircraft, satellite imagery and long treks across the pampa, and still missed a great deal.
That is what makes the new count consequential. The team reported 303 newly identified relief-type, or surface-type, geoglyphs, lifting the confirmed total to 430 such figures, a group that includes 318 the university had recorded since 2004. The additions are not a marginal update to the catalog; they roughly double the known population of this particular class of figure, and they arrived in a fraction of the time manual survey would have required.
The figures themselves lean toward the intimate rather than the monumental. Many are small, averaging around 20 yards in length, and several depict human heads with exaggerated eyes or mouths, along with animals such as camelids and felines that were woven through daily life in ancient Peru. That is a different visual vocabulary from the giant line-type designs that made the site famous.
How the AI found what surveys kept missing
The research team, led by archaeologist Masato Sakai, trained an AI model on already-known geoglyphs, then set it loose on high-resolution aerial imagery covering the plateau, as detailed in the PNAS paper. Rather than searching for whole drawings, the system broke the landscape into thousands of small tiles and checked each for patterns consistent with a possible geoglyph. Researchers also split known figures into smaller sections to expand the training data, which sharpened the model’s ability to flag faint or incomplete shapes.
The AI identified the 303 new surface-type geoglyphs between September 2022 and February 2023, according to the study as summarized in coverage of the work. Human archaeologists then did the confirming, because a machine flag is a candidate, not a conclusion. The role of the model was to narrow an enormous search space, pointing fieldworkers toward the most promising ground.
One pattern jumped out once the figures were mapped. Many of the newly found geoglyphs sit close to old walking routes across the plateau, several within roughly 141 feet of ancient paths. The researchers suggest that travelers may have encountered the images during rituals or used them as visual markers tied to myths or social roles. Sakai has noted that Nazca society had no written language, and proposed that people “learned about the roles of humans and animals by looking at pictures.” Those interpretations remain informed inference, not settled fact, and the study frames them as such.
What it means for the desert and for the discipline
The Nazca Pampa stretches across more than 190 square miles, and combing it on foot is punishing, slow work. An AI that can triage likely locations in days compresses the front end of discovery and lets archaeologists spend their limited time confirming finds and protecting fragile ground. That protection is not incidental: modern roads, mining and tourism continue to threaten the thin desert surface where the geoglyphs survive, and the new digital maps could help authorities decide where to concentrate preservation.
For the wider field, the authors pitch their method as a template, writing that combining field survey with AI establishes “a new paradigm” that makes investigation more efficient. It is worth keeping expectations calibrated. The model did not interpret the figures or explain why the Nazca made them; it accelerated the finding. The meaning of these images, and the culture that left no written record behind them, still rests on human archaeology.
What to watch next is whether the same approach travels. If a model trained on Nazca can be adapted to other arid, image-rich landscapes, the count of known ancient markings could rise quickly elsewhere, which would sharpen an old question: how do you safeguard sites that satellites and algorithms can now find faster than heritage agencies can protect them?
More from Morning Overview
- The NSA issues an urgent warning to all phone users about a setting most people leave on
- 5,000-year-old wolves on remote island upend domestication theory
- Scientists may have found a Type II civilization
- EPA finally kills ‘stupid feature’ in 60% of U.S. cars after 14 years of driver rage
*This article was researched with the help of AI, with human editors creating the final content.
