By Yang Qingyue
Archaeology and artificial intelligence (AI) may seem like two different worlds, but they are coming together in a fascinating way.
Archaeologists dig through the dust to uncover the past, while AI has the computational power to process vast amounts of data. This powerful combination is revolutionizing archaeological research, as AI technologies, with deep learning at its core, are increasingly used to explore and understand human history.
Imagine having a “perspective eye” that can penetrate dense forests and unveil ancient ruins hidden underground – that’s the wonder created by airborne laser scanning technology. Acting as the “X-ray” in archaeology, it can emit lasers from aircraft to scan the ground, even exposing ancient cities nestled deep within jungles with remarkable clarity.
This technology can produce an extensive amount of data, just like capturing an entire city in high-definition imagery. Archaeologists and AI experts leverage deep learning methods like convolutional neural networks to train AI “assistants”so that they can automatically identify ancient ruins from the airborne laser scanning data, significantly boosting the efficiency of archaeological discovery.
Take oracle bone fragments as an example. Chinese oracle bones are pieces of ox scapula or tutle plastron, which were used for divination during the late Shang Dynasty (1600 BC-1046 BC). They bear the earliest significant corpus of ancient Chinese writing and contain important historical information about the Shang. Over 100,000 pieces of oracle bone have been found so far.
Recently, Chinese archaeologists and AI scientists have developed a novel “puzzle-solving tool” – a network model based on internal similarity, which can solve puzzles in three steps.
Finding edges – The model adopts the”edge equidistant matching” algorithm to carefully search along the edges of oracle bonefragments and identify shape-similar parts that might interlock, which is much like finding edge pieces in a jigsaw puzzle.
Piecing together – In the second step, the model will piece together two fragment images to see if they can align perfectly.
Verifying patterns – This is the most critical step, which involves analyzing the consistency of textures at the joints. When there are inscriptions at the joints, it is also important to examine whether these inscriptions are properly aligned. If the two fragments were originally a whole, the texture at the joint should naturally continue.
The model employs a unique technique to accurately measure the similarity of these textures. Experimental results have shown that the model has an accuracy rate of over 90 percent. The magical “puzzle-solving tool”has already helped archaeologists identify 37 pairs of oracle bone fragments that can be correctly assembled.
The Nasca lines in southern Peru, a UNESCO world heritage site, are enigmatic legacies left by ancient civilizations. These massive ground drawings, also known as geoglyphsstretching hundreds of meters, are difficult to discern from the ground and can only be fully appreciated from high above, posing a significant challenge to archaeologists. For years, researchers have relied on laborious manual searches and low-altitude flights for observation and photography, which was inefficient and may lead to missed details.
Over the past century, archaeologists have discovered 430 geoglyphs. Today, the AI-enabled archaeological research has contributed to the discovery efforts, with an additional 303 geoglyphs identified in just six months. The algorithm functions like an experienced “detective,” analyzing high-resolution aerial photographs to outline the shapes of candidate pictures, which are then confirmed by archaeologists. This has greatly enhanced the discovery and research efficiency.
AI is rapidly transforming many fields of study at an unprecedented pace, driving scientific research into a new era. It is reshaping how we conduct scientific research, endowing us with greater insight and predictive power, helping us better understand the past, navigate the present, and address future challenges.
(Yang Qingyue is a researcher with the Institute of Archaeology, Chinese Academy of Social Sciences.)
