Where should archaeologists dig next?

How Artificial Intelligence Is Helping Discover Lost Amazon Secrets. 

Archeological sites, like those in the Amazon, are disappearing faster than we can find them. Can AI help? 

Original article: Where should archaeologists dig next? The winners of this OpenAI contest can tell them. | National Geographic

Aerial view of the Amazonian jungle on a flight from La Paz to Rurrenabaque, 2014. OpenAI has announced the winners of a its 'OpenAI to Z Challenge' that hopes to help archaeologists cut through the rain forest not with machetes, but with machine learning.

Photograph by Matthieu Paley, Nat Geo Image Collection

🌳 Deep in the Amazon Rainforest — hidden under layers of green trees and vines across nine countries and home to hundreds of Indigenous groups today — there may be thousands of ancient sites where people once lived more than 13,000 years ago! These places hold clues about early civilizations: stone tools, paintings, and buried structures.

But there’s one big problem: the Amazon is HUGE (millions of square miles!), remote, and densely covered by jungle, and really hard to explore. Traditional digging or hiking can’t keep up — sites are disappearing faster than they’re discovered due to development, logging, and climate change .That’s where Artificial Intelligence (AI) comes in.

---

🤖 AI to the Rescue!

Two archaeologists teamed up with OpenAI, the company behind ChatGPT, to launch a global contest. The challenge?

➡️ Use computers to scan satellite images and find hidden signs of ancient human activity in the rainforest.

OpenAI let tech experts use Artificial Intelligence (AI) and machine learning to analyze massive sets of satellite images and remote sensing data to spot possible hidden archaeological sites.

The winning team, called Black Bean, used super-smart AI models to analyze huge sets of data — from NASA, Google Earth, and LiDAR scans (special sensors that use laser light to map the land).

🌍 They found 67 possible sites across the Amazon — many near rivers and streams, which makes sense because ancient people often lived near water. Each spot is roughly one square mile and might contain clues to ancient human life.

💬 “Our results make sense!” said Yao Zhao, one of the winners. “Ancient people needed water to live — and that’s where we found clues.”

🏆 The team won $250,000 and special credits to keep using OpenAI’s advanced tools.

 

Team Black Bean 🫘used:

• 🌐 Public satellite images from NASA and Google Earth Engine

• 🛰️ LiDAR data, which uses laser pulses to map ground elevation even through tree cover

• 🧠 GPT-4o, an OpenAI model trained to recognize the patterns of known archaeological sites and apply them to unexplored areas

The team trained deep learning models on several publicly available datasets, including remote sensing LiDAR data and satellite images from Google Earth Engine and NASA’s digital elevation models, among others. 

The team says they then used OpenAI's GPT-4o model to learn the pattern of known archeological sites in the Amazon rainforest and compare them to unexplored swaths of the Amazon, mainly in Brazil. It then highlighted dozens of coordinates for future exploration. 

 

LiDAR satellite image showing the Rio Negro, one of the largest tributary rivers of the Amazon. The winners of the OpenAI archaeology challenge identified 67 distinct square mile patches they recommend explorers search to find potential ancient sites. Many were near water.

Photograph by Matthew Hurst, JAXA/Science Photo Library

---

🛰️ How This Tech Helps Archaeologists

Machines can scan millions of square miles of data in weeks, finding patterns humans might miss — something humans would take years to do! It helps scientists spot patterns and find potential sites faster.

Dr. Sarah Parcak, a space archaeologist, has already used satellite images, thermal imaging and LiDAR to find lost tombs and settlements in Egypt and Tunisia. She says AI is the next big step: 

it can look beyond known sites to reveal completely new areas to explore. With new AI tools, researchers can explore entirely new regions — not just known historical areas.

 

But time is running out. Rising seas, construction, and deforestation are erasing ancient history faster than we can study it.

🕰️ “We have a limited time to document Earth before it changes forever,” says archaeologist Chris Fisher.

---

⚖️ The Ethical Side

Not everyone is cheering. Some critics say companies like OpenAI should have consulted Indigenous communities living in the Amazon first. These groups may see the exploration of their ancestral lands as intrusive or disrespectful.

Brazil’s Indigenous ministry even asked OpenAI to pause the challenge until it could explain its purpose.

OpenAI responded that all data used was publicly available, and no maps included uncontacted Indigenous tribes.

Zhao’s team also said they plan to share results with scientists — but won’t visit any sites without careful consideration for the local communities.

---

🏺 What’s Next?

Parcak and Fisher think more private companies will start using AI to help archaeologists, as government funding for archaeology declines.

Both archaeologists agree that AI won’t replace human experts. Instead, it will help them uncover more of Earth’s hidden history faster.

🧠 “AI scales what remote sensing scientists have been doing for decades,” says Parcak. “It’s not about replacing us — it’s about empowering us.”

But they remind everyone:

“AI won’t replace archaeologists. It just helps us work smarter.”

🧭 The future of archaeology might start with a computer — but it still needs human curiosity to uncover the past.