Google SWE points 48 AI agents at NASA Lidar, finds unmapped Maya pyramid in 72h
First reported by Pietromiotti ·
You can now discover significant archaeological sites with AI agents and lidar data in less than three days.
A software engineer utilized 48 AI agents to analyze NASA lidar data and discover a previously unmapped Maya archaeological site in Mexico. The process began with a casual thought about historical explorers and modern tools, leading to the exploration of lidar data, which can penetrate jungle canopies to map the ground. The AI agents processed approximately 1,200 km² of lidar strips, identifying potential structures like pyramids, platforms, and plazas by visually inspecting generated images. One significant flagged site on a cliff edge appeared to be a small Maya civic center from the Classic period, complete with a pyramid, plaza, and residential platforms. The engineer then compiled a report with the agents' assistance, translated it into Spanish, and submitted it to Mexico's National Institute of Anthropology and History (INAH). INAH confirmed that no registered archaeological site existed at the reported coordinates, marking a potentially significant new discovery made in under 72 hours.
This discovery signals a paradigm shift in how scientific research, particularly in archaeology and potentially other earth sciences, can be accelerated through the application of generative AI. The ability of AI agents to independently sift through vast datasets like lidar, identify patterns, and flag anomalies, drastically reduces the time and human effort traditionally required for such tasks. It democratizes access to complex data analysis, potentially enabling individuals without specialized domain knowledge to make novel scientific contributions.
The rapid identification of a new Maya site by an independent researcher, powered by AI, suggests that vast swathes of the planet may still hold undiscovered historical and geographical information accessible through similar methods. This could trigger a new wave of discovery across various scientific disciplines, challenging existing inventories and potentially rewriting historical timelines. Future research will likely focus on scaling these AI-driven analysis techniques to global lidar datasets and integrating them into formal scientific workflows.
AI-written summary. May contain errors.