A Google researcher, Pietro Miotti, announced on November 10th that he had located an unmapped Mayan archaeological site in Mexico using 48 artificial intelligence agents. The idea arose during a trip to Peru, when Miotti began to speculate on what past explorers would do with modern technologies. Without an archaeology background, he created a personal project to analyze public LiDAR data, a three-dimensional laser mapping technique provided by NASA.
Due to the limitation of public data coverage in Peru, the researcher focused the algorithms on the Mayan region. The AI agents identified several interesting points, including what appeared to be a large village without official registration. Miotti used the tools himself to translate the communication into Spanish and consulted the National Institute of Anthropology and History of Mexico (INAH), which confirmed having no registered records at those coordinates.
Excited by the response in less than 72 hours, Miotti publicized the achievement on his blog, declaring that humanity was on the verge of a scientific revolution and advocating for global LiDAR mapping to reveal hidden history. He also posted about it on his X account. Despite mentioning that the finding would require fieldwork for confirmation, one specialist would quickly notice a problem: the site was already known to the scientific community and had been published in an academic journal over a year ago.
The archaeological community quickly pointed out the flaw on social media. Sarah Parcak, an archaeology professor at the University of Alabama in Birmingham, emphasized the importance of peer review and comparing new data with existing scientific literature. She criticized the project for spreading misinformation and diminishing researchers who have been using advanced digital tools for decades, arguing that the absence of a record in the official Mexican database does not imply ignorance.
Elizabeth Graham, a Mesoamerican archaeology professor at University College London, also contested the idea of 'discovery,' reminding that local populations often already know these vestiges. For her, the main obstacle in archaeology today is not finding new sites, as there are many foot-identified sites awaiting study, but rather the lack of funding for excavation and preservation of what has already been discovered.
In response to the criticism, Miotti admitted his mistake to The Verge, explaining that because he is not an archaeologist, he did not know that the lack of government registration did not mean the site was unprecedented. He justified that his agents failed to cross-reference the information because, for security reasons, scientific articles do not disclose exact coordinates, requiring visual comparison of maps and figures, something models still struggle to do. After being warned, the researcher updated his publications to give credit to the original work led by Francisco Estrada-Belli of Tulane University, while maintaining the belief that the workflow can help specialists filter large volumes of data.
This incident reveals a growing conflict in modern science: the enthusiasm of technology users and professionals who, armed with powerful tools, enter highly specialized areas without mastering the academic context. As seen in other fields, such as mathematics, AI systems can produce remarkable results but often fail to dialogue with the prior knowledge that underpinned their 'discoveries,' claiming advancements that are already established and shifting the exhaustive work of verifying true scientific value onto specialists.
Ivan Šprajc, an archaeology professor at the Slovenian Academy of Sciences and Arts and co-author of the original study documenting the Mayan site, stated he was not impressed by the feat. He recalled that archaeologists have been using LiDAR and remote sensing for years, and that merely pointing out an anomaly on the ground does not generate knowledge without proper historical interpretation. Šprajc also warned that the mass submission of supposed 'discoveries' made by amateurs threatens to overwhelm researchers, reporting having already received similar emails from other AI users.
Beyond the academic noise, the premature disclosure of archaeological findings carries serious practical risks. By publicly exposing the location of ancient ruins and settlements without adequate institutional support or preservation plans, enthusiasts end up providing a direct map for heritage plunderers, turning a tool supporting science into a real threat to the conservation of vulnerable sites.


