Google AI researcher Pietro Miotti said his AI agents found an unmapped Maya site under the jungle canopy in Mexico, but the site was already in the archaeological record. The agents just missed the research documenting it.
Miotti had made a different claim in his original blog entry: “By analyzing NASA lidar data, my AI agents discovered a Maya site in Mexico that, as far as anyone can tell, no one had reported before.” He also highlighted the speed of the authorities’ response: “It took less than 72 hours to go from a random thought while traveling to an official reply from the Mexican government.”
Although Miotti acknowledged that he’s “not an archeologist,” he treated the apparent find as evidence of a wider shift. “This is by far the best time ever to be alive,” he wrote. On X, he expanded on that view: “I think we’re on the brink of a revolution across every science, not just coding, physics and math.”
The claim soon spread across social media. A user praised Miotti for “finding ancient mayan sites from your desk chair,” while another used the apparent discovery to defend AI data centers. Posts also circulated AI-generated pictures imagining the site after excavation. Both The Verge and TechRadar covered the story.
Archaeologist and award-winning author Sarah Parcak soon challenged the idea that this was a discovery. “This story is going wild on X and elsewhere,” she wrote in a Bluesky response. Her correction was direct: “The site was already published by Dr. Francisco Estrada-Belli of Tulane University ages ago.”
For a paper published in 2025, Estrada-Belli’s team had studied lidar data collected by NASA’s Goddard Space Flight Center across 95,000 square kilometers of Central America. Their findings suggested that previous estimates may have substantially understated the ancient Maya population, which may have exceeded 16 million people at its height.
Parcak distinguished a reply from officials from evidence of a new discovery. “This is why we have peer review. Mexican gov didn’t have it in their database, but that doesn’t mean it is unknown/unmapped.” Estrada-Belli subsequently provided further support for her account by sharing images of the site with her.
Miotti later revised his post, acknowledging that “it turns out the site had already been mapped.” He attributed the mistake to the literature search: “The agents missed it while going through the literature, so they rediscovered it rather than discovered it.” His closing comment was: “Still work to do then!”
Spotting structures in lidar imagery was only half the job. Before calling something a discovery, you have to check published research to see whether the site is new, and the agents failed at that step.
The episode joins other cases in which AI researchers have faced accusations of claiming credit for scientific work that already exists. OpenAI’s publication of numerous “proofs” addressing well-known, unresolved mathematics problems is perhaps the strongest example. Reports about those solutions suggested that many improperly drew on work by mathematicians who were still tackling the problems.
That conflict became a crisis earlier in 2026, with mathematicians accusing OpenAI of offering unverified proofs and relying on plagiarism. Futurism ties those complaints to a larger criticism: that the AI industry built its foundations by effectively ripping off rights-protected material from the web to train its models. Miotti’s case doesn’t establish those broader allegations, but it raises a related problem. He corrected his claim, but similar assertions by other researchers could be taken as fact without anyone challenging them.
These narratives can also obscure how long specialists have worked with such technology. Geoscience consultant Alex Janevski noted on Bluesky that machine learning and computer vision have been a “core part of remote sensing for as long as it’s been around.”
For Roman historian Anise Strong, the problem concerned both misallocated credit and a misleading picture of experts’ methods. “I hate how this sort of hyperbolic misinformation both takes credit away from the actual experts and distorts how those experts are of course already using advanced digital tools of various kinds and have been for decades,” she said on Bluesky. “Don’t reinvent the wheel, Random Internet Guy!”
