AI Agents ‘Discover’ Maya Ruins Already Documented by Archeologists

Earlier this week, Pietro Miotti, a researcher at Google, believed he had used artificial intelligence to achieve what explorers of the past could only dream of: discovering a lost Maya settlement in the Mexican jungle. However, the "discovery" has instead become a cautionary tale about the limits of AI when stripped of domain expertise.

While Miotti’s team of 48 AI agents successfully flagged a site in Mexico, archeologists were quick to point out that the location was already known and had been published in an academic journal just last year.

48 Agents and NASA Data

The project began as a personal side venture while Miotti was traveling in Peru. Inspired by the potential of modern technology to mirror historic exploration, he turned to lidar (laser-based mapping technology) data provided by NASA. Lidar is capable of producing high-resolution 3D images of the ground, even through dense forest canopy.

Because public lidar data for Peru is limited, Miotti shifted his focus to the Maya region. He deployed 48 AI agents to scan the datasets for regular geometric shapes that might indicate human-made structures. The results were initially stunning.

“I woke up and checked the agents’ reports. ‘DAMN!’ They had found lots of villages and other regular shapes that looked like structures or houses,” Miotti wrote on his blog. One specific location appeared to be a major village that was missing from existing public records.

The "72-Hour" Discovery

To verify his findings, Miotti used his AI agents to translate correspondence into Spanish and reached out to Mexico’s National Institute of Anthropology and History (INAH). According to Miotti, the institute responded that they had no record of a registered site at those specific coordinates.

“It took less than 72 hours to go from a random thought while traveling to an official reply from the Mexican government,” Miotti noted, highlighting the speed at which AI can process information and facilitate communication.

The Expertise Gap

The excitement was short-lived. Once the claim reached the broader archeological community online, experts pointed out that the site was not "undiscovered." It had been documented and published in a peer-reviewed journal in 2025.

The incident has sparked a heated debate regarding "amateur parachuting"—a phenomenon where individuals use powerful AI tools to enter specialized fields without the necessary background knowledge. Archeologists expressed frustration that AI-driven "discoveries" often ignore existing academic literature, unpublished research, and the knowledge of local communities who are often well aware of the sites in their own backyards.

While Miotti clarified that the research was a personal project conducted without Google’s computing resources, the situation underscores a growing tension in the age of AI. As tools become more accessible, the distinction between a genuine scientific breakthrough and the rediscovery of known facts becomes increasingly blurred.