AI
Google researcher's "AI-discovered" Maya site was already published last year
An amateur scientist used AI agents to analyze lidar data and claim a discovery in Mexico—only to learn the site had been documented in academic literature more than a year prior.
1/2
The Verge
A Google researcher recently claimed success in using a network of AI agents to uncover a previously unknown Maya archaeological site in Mexico. The catch: specialists had already identified and published research on the location over a year ago. The mishap has prompted growing frustration from the academic archaeology community about nonprofessionals deploying powerful AI tools in specialized fields without the domain knowledge to contextualize their findings.
The Claimed Discovery#
Pietro Miotti recounted how inspiration struck while traveling in Peru—he wondered how historical explorers might have benefited from modern technology. He chose lidar, a laser-based sensor system capable of generating precise 3D environmental maps, and deployed 48 AI agents to sift through NASA's publicly available lidar datasets in search of undocumented structures. After shifting his focus to Maya regions (due to limited Peru coverage), the agents identified multiple candidate locations, flagging one "major village" that appeared absent from official registries.
Miotti then contacted Mexico's National Institute of Anthropology and History (INAH) to verify whether the site was already catalogued. When INAH indicated no official record existed, he announced the findings publicly. "It took less than 72 hours to go from a random thought while traveling to an official reply from the Mexican government," he wrote, calling the result proof that AI was revolutionizing scientific discovery.
The Reality Check#
Archaeologists quickly identified a fundamental problem: the site had been documented and published in a peer-reviewed journal more than a year earlier. Sarah Parcak, an archaeology professor at the University of Alabama at Birmingham, and other specialists highlighted on social media that absence from a government database does not equate to being unknown or unmapped. As Parcak noted, this underscores why peer review exists—and why checking existing literature is essential before claiming discoveries.
Deeper Issues at Stake#
The incident exposes several layers of difficulty. Government agencies don't maintain complete inventories of all known sites, particularly those explored by researchers or known to local communities. Archaeologists have already been employing lidar and other remote sensing technologies for years—the added value of AI flagging pixels isn't groundbreaking on its own. Ivan Šprajc, an archaeologist who co-authored the paper documenting this particular site, observed that "just spotting something does not add much to our knowledge" without proper contextual understanding and fieldwork.

