Google Engineer Uses AI Agents to Discover Unmapped Maya Pyramid
Forty-eight AI agents analyzed NASA lidar data to identify an ancient site in Mexico.
1Rapid Archaeological Discovery
A Google software engineer utilized 48 AI agents to identify a previously unrecorded Maya site in Mexico. By processing public NASA lidar data, the system successfully mapped a hilltop structure within 72 hours. This process transitioned from an initial concept to an official confirmation from the Mexican government in record time.
2Lidar Data Processing
Lidar technology fires laser pulses to penetrate dense jungle canopies, mapping the ground surface beneath. The agents converted 1,200 square kilometers of lidar strips into visual images. By systematically filtering out natural terrain features like sinkholes and hills, the agents identified man-made structures including a plaza and an oval pyramid.
3Agentic Workflow Execution
The project employed multiple models, including Gemini, ChatGPT, Claude, and Grok, to perform the analysis. Rather than using specialized classifiers, the agents followed a precise visual guide to identify straight edges and sharp corners. They cross-referenced findings against literature and archives to ensure the site had not been previously registered.
4Site Structural Analysis
The identified site features a levelled plaza, an 8–12 meter tall pyramid, and a raised courtyard. A flat-bottomed hollow suggests a potential aguada, or rainwater reservoir, common in Classic period Maya towns. These structures likely served as a civic center for local elites between AD 250 and 900, according to agent-led analysis.
5Official Verification Process
The researcher contacted Mexico's National Institute of Anthropology and History (INAH) with a formal report. INAH confirmed that no archaeological site was previously registered at the specific coordinates provided. While the discovery remains unofficial pending a field inspection, the institute now holds the authority to authorize potential future ground-based research.
6Historical Context of Lidar
Archaeologists have increasingly used lidar to reveal hidden structures, such as the 2024 discovery of an entire Maya city. Previous studies have utilized NASA’s G-LiHT instrument to estimate population density in Maya lowlands. This project builds upon established methodologies where environmental research data is repurposed for significant archaeological and historical mapping efforts.
7Democratizing Scientific Discovery
The project highlights how non-experts can leverage computational power to solve complex research problems. By treating the archaeological search as a data-processing task, the engineer bypassed traditional bureaucratic hurdles. This approach suggests that curiosity and imagination, supported by sufficient compute, are becoming the primary drivers for future breakthroughs in various scientific fields.
8Future of Agentic AI
This experiment demonstrates that agentic AI is applicable beyond coding and physics, offering utility in diverse research domains. The author proposes that lidar scanning the entire planet could lead to massive discoveries. As AI agents become more capable, the bottleneck for scientific progress shifts toward the human ability to formulate questions.
Original article: Hacker News — Google Engineer Uses AI Agents to Discover Unmapped Maya Pyramid