AI agents found a Maya site, but it was a rediscovery, not a discovery

An AI enthusiast set out with a question inspired by Hiram Bingham, the explorer who 'discovered' Machu Picchu in 1911: 'What would Bingham do with our tools?' The user suggested that AI agents scan Maya region using lidar data.

The agents used NASA's open lidar data to convert about 1,200 square kilometers of jungle into images. Forty-eight agents then examined these images one by one, like patient archaeologists, looking for pyramids, platforms, and plazas. According to the report, no special model or classifier was involved; the agents simply looked at pictures and followed a detailed guide.

They found a hilltop with straight edges and sharp corners—features nature rarely creates. Zooming in revealed a leveled plaza, an oval pyramid at least 8–12 meters tall, and a flat-bottomed hollow that might have been a water reservoir. When the agents realized the site wasn't in the literature or archives, they advised the user to contact INAH, Mexico's National Institute of Anthropology and History. INAH replied that there was no registered archaeological site at that location.

But the twist came soon after: the site had already been mapped in a 2025 paper by Estrada-Belli et al. The article notes that the agents missed this during their literature review, so they rediscovered it rather than discovered it. Still, the site wasn't yet in INAH's official register. This shows how quickly AI agents can produce results in archaeology, but also that careful verification is essential.