Key points
- Pietro Miotti, a Google researcher, used AI agents to analyse NASA lidar data for archaeological sites.
- The agents flagged a location in Mexico that Miotti believed was unregistered.
- The site had been discovered and published in an academic journal the previous year.
A Google researcher claimed this week that AI agents identified a potentially unmapped Maya site in Mexico. The claim drew immediate criticism because the location had already been documented in an academic journal last year.The site in question had already been found and written about in a scientific journal last year.The failure of AI agents to identify an existing academic publication underscores the limitations of automated discovery in specialist domains.
The evidence
Google researcher Pietro Miotti deployed 48 AI agents to scan NASA lidar data for evidence of unrecorded sites.reported He described the effort as a personal side project that did not use Google compute or resources. The AI agents identified several potential sites, among them a large village that was not listed in current records. Miotti contacted Mexico’s National Institute of Anthropology and History (INAH) to determine if the site was already known. He used AI to translate his messages into Spanish and stated that INAH indicated it had no record of a registered site at that location.
The decision
Miotti posted on his blog that the process, from a spontaneous idea during travel to receiving an official response from the Mexican government, took under 72 hours.reported Miotti stressed that he is not an archeologist and emphasized that the finding would need to be verified with fieldwork. The incident sparked criticism among archaeologists online, underscoring a growing tension in science. Scientists are reportedly weary of non-specialists entering their disciplines and claiming AI-driven discoveries.
What to make of it
The incident highlights the limitations of AI agents when applied to specialist fields without domain expertise. While visiting Peru, Miotti conceived the notion of employing AI to locate archaeological sites. He pondered how past explorers would act if they had modern technology.
In practice
For users of AI agents, this case demonstrates that automated tools can miss existing public records. Miotti’s agents failed to identify that the site was already published in an academic journal last year. Local communities are often already well aware of sites in their area. Users must verify AI-generated findings against established academic and local sources. The reliance on AI for translation and data analysis does not guarantee accuracy in complex, specialist domains.
Why it matters
- AI agents may overlook existing public records, leading to false claims of discovery.
- Lack of domain expertise can result in misinterpretation of data, even with advanced tools.
- Automated workflows require human verification to avoid duplicating known findings.
- The incident highlights the tension between amateur AI users and professional experts.
Topics: AI agents, lidar, archaeology, data analysis, Google
Updated
Then / now / next
Archaeological discoveries were primarily made by experts with domain-specific knowledge and fieldwork experience.
Amateurs are using AI agents to analyse public data, sometimes claiming discoveries that are already known.
Watch for how academic institutions and AI developers address the verification of AI-generated findings in specialist fields.
Story timeline
- Then Archaeological discoveries were primarily made by experts with domain-specific knowledge and fieldwork experience.
- 12 Nov 2026 No, AI agents did not discover a previously unknown Maya site
- Now Amateurs are using AI agents to analyse public data, sometimes claiming discoveries that are already known.
- Next Watch for how academic institutions and AI developers address the verification of AI-generated findings in specialist fields.
For readers
Researchers
- AI tools can miss existing academic publications, leading to redundant or false claims.
- Domain expertise is essential for interpreting data generated by automated systems.
Developers
- Automated agents may fail to account for prior knowledge or local awareness.
- Verification steps are necessary to ensure the accuracy of AI-generated findings.
Policy
- The incident highlights the need for guidelines on amateur use of AI in specialist fields.
- Tension between amateurs and experts may require new standards for validation.
Ask the article
Local keyword search of this article; no generated answers.Did this article give you the evidence you needed?
Your choice stays on this device. No public tally.Questions readers ask
Did AI agents discover a new Maya site?
No, the site flagged by the AI agents had already been discovered and published in an academic journal last year.
Who conducted the AI search for the Maya site?
Pietro Miotti, a Google researcher, conducted the search as a personal side project using NASA lidar data.
What was the response from Mexican authorities?
Miotti stated that Mexico’s National Institute of Anthropology and History (INAH) informed him there was no record of a registered site at the location.
Connected stories
Reading receipt
- Pietro Miotti, a Google researcher, used AI agents to analyse NASA lidar data for archaeological sites.
- The agents flagged a location in Mexico that Miotti believed was unregistered.
- The site had been discovered and published in an academic journal the previous year.
Still open: How can AI tools be improved to better account for existing academic records?
Still open: What guidelines should be established for amateur use of AI in specialist fields?