Lost Worlds Found by AI
AI is giving archaeologists new sight. In central Mesopotamia, deep learning models retrained on CORONA satellite photographs—grainy black-and-white images captured by U.S. spy satellites in the 1960s—revealed ancient settlement mounds erased by modern sprawl. These Cold War archives preserve landscapes before tractors and highways leveled them. By teaching convolutional networks to detect faint traces of tells, researchers mapped 81 sites, many long destroyed, and confirmed four unknown settlements through field surveys. Machines are now restoring memory, retrieving lost worlds that progress once buried.
Further Reading:
PLOS One Study → | arXiv Preprint →
Austen Henry Layard, Excavations at Nineveh, 1852. Public domain.