In a discovery that’s turning heads across the archaeological world, researchers have used artificial intelligence to uncover 303 previously unknown Nazca geoglyphs in the Peruvian desert, nearly doubling the number of documented ancient figures etched into the arid landscape.

The findings, detailed in a peer-reviewed study published in PNAS, mark a major leap forward in the study of the enigmatic Nazca culture and suggest a far more complex ceremonial and social use of these sprawling ground drawings than previously thought.

The project, a collaboration between Yamagata University in Japan and IBM Research, relied on deep learning to scan over 629 square kilometers of high-resolution aerial and drone imagery. The AI system, trained on a relatively small dataset of known geoglyphs, was able to detect faint, shallow, and weathered relief-type figures—many as small as 9 meters across—that have eluded human researchers for decades.

“This technology has allowed us to condense nearly a century of archaeological progress into just six months,” said Professor Masato Sakai, lead archaeologist at Yamagata’s Institute of Nazca.

The Overlooked Geoglyphs That Reshaped Archaeological Thinking

Unlike the more famous line-type Nazca geoglyphs—large stylized animals like monkeys, hummingbirds, and whales that stretch up to 90 meters and were first studied from the air in the early 20th century—the newly discovered figures belong mostly to the lesser-known relief-type category.

These smaller figures, meticulously outlined by removing surface stones to expose the lighter earth beneath, depict a range of human-related motifs: humanoids, decapitated heads, and domesticated animals like camelids. In fact, over 80% of the new finds depict human-modified subjects, in stark contrast to the wildlife-centric themes of the larger geoglyphs.

Crucially, these relief-type geoglyphs are often located within 43 meters of ancient foot trails, suggesting they were designed to be viewed by individuals or small groups traveling across the Nazca Pampa—not by aerial observers or large congregations. This supports earlier hypotheses proposed by German mathematician and Nazca researcher Maria Reiche, who posited that many geoglyphs were tied to ritual processions.

By contrast, the massive line-type figures tend to cluster around linear and trapezoidal paths, believed to be part of community-wide ceremonial networks. These findings lend weight to the idea that Nazca geoglyphs served a dual-purpose landscape: intimate, localized rituals and broader, communal pilgrimage activity.

AI’s Role in Rewriting Ancient Narratives

The AI’s success in detecting such difficult-to-spot figures came down to clever engineering and a bit of patience. Because of the limited training data—just over 400 known geoglyphs at the time—researchers fine-tuned a model pre-trained on conventional photographs, enhancing it with custom algorithms that scanned the imagery in 5-meter grids. A geoglyph probability map was then generated, helping archaeologists prioritize field surveys.

The team manually examined over 47,000 AI-flagged image boxes, spending more than 2,600 labor hours on screening and field verification. The payoff was significant: 303 new figurative geoglyphs confirmed between September 2022 and February 2023, alongside 42 new geometric figures and dozens of new groupings not previously documented.

This approach also revealed that many geoglyphs cluster in narrative scenes—for example, humanoids interacting with animals or symbolic decapitation motifs—further supporting the idea that the Nazca used these trails and figures to transmit cultural memory and ritual significance through motion and space.

“AI doesn’t replace the archaeologist,” said Dr. Alexandra Karamitrou, an AI researcher at the University of Southampton not involved in the study. “But it radically expands what’s possible, especially in places as vast and harsh as the Peruvian desert.”

Cultural Heritage Under Threat and a Race Against Time

This technological advance comes at a pivotal moment. The Nazca geoglyphs, designated a UNESCO World Heritage Site, face growing threats from climate change, unauthorized vehicle incursions, and flash flooding—phenomena becoming more frequent in the desert due to shifting weather patterns.

Preserving these fragile expressions of ancient Andean culture is now as much about data as it is about dirt. The AI-assisted survey not only improves the mapping of known figures but also highlights potential hot spots for future discoveries, many of which lie just beneath the surface of satellite scans.

With roughly 1,000 AI-flagged candidate sites still awaiting verification and many trails only partially mapped, researchers expect hundreds more figures may remain undiscovered. If so, we’re only beginning to grasp the cultural sophistication of a civilization that, over 1,500 years ago, etched stories into stone—not for us, but for the gods, the landscape, and each other.

Artificial Intelligence is everywhere, and we seemingly can’t escape.

I’ve never (and will never) use AI to write articles on Windows Central, beyond perhaps using Copilot to quickly check the specs on a product I’m reviewing — but even that often requires additional review, due to the hallucinations AI seems prone to. It seems like we might be increasingly in the minority, though.

Week 2 of Sunday Night Football will see the Minnesota Vikings (1-0) hosting the Atlanta Falcons (0-1). J.J. McCarthy and Michael Penix Jr. will be popular in NFL props, as the two will face off for the first time since squaring off in the 2023 CFP National Title Game. The cast of characters around them has changed since McCarthy and Michigan prevailed over Washington, as the likes of Bijan Robinson, Justin Jefferson and Aaron Jones now flank the quarterbacks. There are several NFL player props one could target for these star players, or you may find value in going after under-the-radar options.

Tyler Allgeier had 10 carries in Week 1, which were just two fewer than Robinson, with the latter being more involved in the passing game with six receptions. If Allgeier has a similar type of volume going forward, then the over for his rushing yards NFL prop may be one to consider. A strong run game would certainly help out a young quarterback like Penix, so both Allgeier and Robinson have intriguing Sunday Night Football props. Before betting any Falcons vs. Vikings props for Sunday Night Football, you need to see the Vikings vs. Falcons prop predictions powered by SportsLine’s Machine Learning Model AI.

Built using cutting-edge artificial intelligence and machine learning techniques by SportsLine’s Data Science team, AI Predictions and AI Ratings are generated for each player prop. 

For Falcons vs. Vikings NFL betting on Sunday Night Football, the Machine Learning Model has evaluated the NFL player prop odds and provided Vikings vs. Falcons prop picks. You can only see the Machine Learning Model player prop predictions for Atlanta vs. Minnesota here.

Top NFL player prop bets for Falcons vs. Vikings

After analyzing the Vikings vs. Falcons props and examining the dozens of NFL player prop markets, the SportsLine’s Machine Learning Model says Falcons RB Bijan Robinson goes Over 65.5 rushing yards (-114 at FanDuel). Robinson ran for 92 yards and a touchdown in Week 14 of last season versus Minnesota, despite the Vikings having the league’s No. 2 run defense a year ago. After replacing their entire starting defensive line in the offseason, it doesn’t appear the Vikings are as stout on the ground. They allowed 119 rushing yards in Week 1, which is more than they gave up in all but four games a year ago.

Robinson is coming off a season with 1,454 rushing yards, which ranked third in the NFL. He averaged 85.6 yards per game, and not only has he eclipsed 65.5 yards in six of his last seven games, but he’s had at least 90 yards on the ground in those six games. Over Minnesota’s last eight games, including the postseason, six different running backs have gone over 65.5 rushing yards, as the SportsLine Machine Learning Model projects Robinson to have 81.8 yards in a 4.5-star prop pick. See more NFL props here, and new users can also target the FanDuel promo code, which offers new users $300 in bonus bets if their first $5 bet wins:

How to make NFL player prop bets for Minnesota vs. Atlanta

In addition, the SportsLine Machine Learning Model says another star sails past his total and has five additional NFL props that are rated four stars or better. You need to see the Machine Learning Model analysis before making any Falcons vs. Vikings prop bets for Sunday Night Football.

Which Vikings vs. Falcons prop bets should you target for Sunday Night Football? Visit SportsLine now to see the top Falcons vs. Vikings props, all from the SportsLine Machine Learning Model.