How AI Has Started a Fire in Archaeology

Instead of being obsessed about the future, what if we used AI to learn from the past and unlock its secrets?

As big tech continues to hoover up as much data as possible to feed the AI machine, experts believe we could run out of high-quality data by 2026.

Yet if we point the telescope in the other direction of time, vast quantities of data are just waiting for AI to bring them back to life and reveal stories of human innovation from a million years ago.

We learned recently how Israeli researchers are using sophisticated artificial intelligence techniques to dig up proof of the oldest fire on earth — 800,000 years ago, at least.

So we donned our Indiana Jones replica hats and went on an AI archaeology adventure to see what was cooking in the first instances of our ancestors.

Table of Contents Table of Contents

1. Key Takeaways
2. The Role of AI in Archaeology
3. How AI and Spectroscopy Uncover Ancient Fire Sites from 800,000 Years Ago
4. The Bottom Line
5. References

Show Full Guide

Table of Contents

1. Key Takeaways
2. The Role of AI in Archaeology
3. How AI and Spectroscopy Uncover Ancient Fire Sites from 800,000 Years Ago
Show Full Guide
4. The Bottom Line
5. References

The Role of AI in Archaeology

Many people don’t associate archaeology with high-profile tech projects when thinking about the past. But Dr. Filipe Natalio and Dr. Ido Azuri from the Weizmann Institute of Science see things differently.

In a recent conversation, Natalio emphasized to me the crucial role of data in maximizing the use of AI developer methods, and they had an almost unlimited supply.

“You need data. So, we are talking about hundreds, probably millions of years of cultural material that has been produced.

“Museums have hundreds of thousands of things that have been generated by humans throughout this time. So there is no shortage of large databases that you could use.”

Ancient fire use detection is important because it is considered a key factor in human evolution. AI models facilitate the detection of sites where visible traces may not exist — opening up new perspectives in archaeology to examine how early humans innovated and evolved.

Debates around the “cooking hypothesis” suggest a range of evolutionary developments, from diet changes to social behavior. When humans gained control over fire, we developed new methods for warming ourselves up, fashioning tools, holding communal gatherings, protecting ourselves from predators, and cooking meat.

“Fire enabled people to move away from the center of their food source. It would also develop bone structure, brain, and cognitive and spatial recognition patterns so humans could walk and explore beyond their food source.”

The oldest evidence of controlled fire was found at Israel’s Qesem Cave around 300,000 to 400,000 years ago. But burnt flint tools found at the Evron Quarry archaeological site have, until now, been hiding secrets that would change our understanding of human industry.

“The site was discovered back in the 70s and was excavated. But now it’s a dump, so we cannot excavate that anymore. When we started this project, all the archaeologists said, ‘You guys won’t find anything because there’s no clear evidence of fire’. And I thought, whoa, that’s a good challenge.”

How AI and Spectroscopy Uncover Ancient Fire Sites from 800,000 Years Ago

The team turned to spectroscopy to measure light absorption in flint artifacts and identify burning evidence. It soon became clear that the data’s output consisted of tiny patterns and variations that were too complex for humans.

The team set out to create an AI program focused on spectroscopic data analysis. The model was designed to classify burned and unburned flint and even estimate what temperatures the flint had been exposed to.

Now, the researchers applied it to 26 flint tools found at the Evron Quarry site. The results were remarkable: the AI picked out heat exposure in the tools, detecting a broad range of heating temperatures reaching as high as 600°C.

This gave them a breakthrough in gleaning meaningful information from data previously impervious to them: Thermal alternations on the rocks revealed the controlled use of fire at different temperatures.

A combination of AI and spectroscopy has proven to be a very powerful tool in unlocking secrets hidden within complex archaeological data. At the Evron Quarry site, using AI to identify patterns and information too subtle for human perception, the researchers were able to provide compelling evidence for possible controlled fire use by early humans.

What is special about this is that only a few archaeological sites with fire are discovered once every decade, and evidence for fire is found only at five archaeological sites.

The AI-assisted findings were evidence of fire use nearly 1 million years ago, and it is one of the oldest known sites outside Africa. This suggests that human ancestors had fire-use capabilities when migrating from Africa earlier than assumed.

One of the biggest challenges in AI applications within archaeology is validating findings due to the time depth and the inability to confirm results with traditional experiments. Felipe and Ido’s team’s interdisciplinary nature, combining chemistry, physics, and AI, highlights the difficulty in interpreting complex data that intertwines human behaviour with archaeological findings.

The Bottom Line

It’s the present moment where technology and historical science meet. By integrating AI and machine learning in archaeology, they are challenging traditional methodologies, and recent evidence could rewrite historical narratives about early human societies

The ability of AI to analyze and interpret large datasets may lead to revelations about ancient human behaviours, technologies, and migrations that give much richer and more detailed information from our past than ever before.