Google researcher’s AI Maya site claim was already published in academic journal
A Google engineer’s AI-driven discovery of a Maya site was debunked after archaeologists noted the location had been documented last year.
Key points
- Pietro Miotti used AI agents to identify a Maya site that was already published in an academic journal.
- The researcher contacted Mexico’s national institute, which stated it had no record of the specific location.
- Archaeologists criticised the incident as an example of amateurs using AI without sufficient domain expertise.
A Google researcher’s claim to have discovered an unmapped Maya site using artificial intelligence was incorrect, as the location had already been identified and published in an academic journal last year, according to the published reportreported
How we got here
Pietro Miotti, a researcher at Google, initiated a personal side project to search for archaeological sites using lidar data, without using Google’s computational resources, The Verge reported.reported He deployed a team of 48 AI agents to analyse NASA lidar data, focusing on the Maya region after finding limited coverage for Peru.reported
What is new
The AI agents flagged several locations, including one major village that Miotti believed was unregistered. He contacted Mexico’s National Institute of Anthropology and History to verify the finding, using AI to translate his correspondence into Spanish.
What to watch
Miotti stated that the institute replied within 72 hours, confirming it had no record of a registered site at that specific location.reported He wrote on his blog that the speed of the process demonstrated the benefits of modern technology, stating it was the best time to be alive.
Availability and limits
For developers, this case illustrates the limitations of AI agents in specialist fields. Miotti emphasised that he is not an archaeologist and that the finding requires fieldwork verification. The incident highlights that AI can process data but may lack the contextual expertise to distinguish between known and unknown entities.
Why it matters
- Developers must recognise that AI agents can produce false positives when operating outside their training context or without specialist validation.
- The incident demonstrates the importance of verifying AI-generated findings against existing academic literature and official records before publication.
- It underscores the risk of amateurs wielding powerful AI tools in specialist fields without the necessary expertise to understand what they are looking at.
- The case highlights a growing tension in science as non-experts use AI to parachute into established disciplines.
Topics: AI agents, archaeology, lidar, data verification, Google
Updated
Then / now / next
AI agents were used to scan lidar data for potential archaeological sites in Mexico.
The identified site was confirmed to have been published in an academic journal last year.
Miotti emphasised that the finding would need to be verified with fieldwork, despite the prior publication.
Story timeline
- Then AI agents were used to scan lidar data for potential archaeological sites in Mexico.
- 12 Nov 2026 No, AI agents did not discover a previously unknown Maya site
- Now The identified site was confirmed to have been published in an academic journal last year.
- Next Miotti emphasised that the finding would need to be verified with fieldwork, despite the prior publication.
The strongest doubt
Counterpoint from the sources
Archaeologists have criticised the incident, noting that local communities and experts are often already aware of such sites, highlighting a tension between amateur AI use and specialist fields.
For readers
Researchers
- Verify AI-generated findings against existing academic literature before claiming new discoveries.
- Recognise that AI tools may lack the contextual expertise to distinguish known from unknown entities.
Developers
- Design systems that flag potential false positives when operating in specialist domains.
- Ensure AI agents are used as aids rather than autonomous authorities in fields requiring deep expertise.
Policy
- Consider guidelines for the use of AI in scientific fields to prevent friction with established communities.
- Encourage collaboration between AI developers and domain experts to validate findings.
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.Terms in this story
lidar
A laser-based mapping technology that produces detailed three-dimensional images of the environment.
AI agents
Automated systems capable of performing tasks or analysing data, such as scanning for patterns in lidar imagery.
Questions readers ask
Did the AI actually discover a new Maya site?
No, the site identified by the AI agents had already been discovered and published in an academic journal the previous year.
What tools did the researcher use?
Miotti used a team of 48 AI agents to analyse NASA lidar data, claiming the project was a personal side project not using Google resources.
How did the researcher verify the finding?
He contacted Mexico’s National Institute of Anthropology and History, which stated it had no record of a registered site at that location.
Connected stories
Reading receipt
- Pietro Miotti used AI agents to identify a Maya site that was already published in an academic journal.
- The researcher contacted Mexico’s national institute, which stated it had no record of the specific location.
- Archaeologists criticised the incident as an example of amateurs using AI without sufficient domain expertise.