Pietro Miotti says he used 48 AI agents to scan images made from publicly available lidar data, which maps ground beneath forest cover. They flagged what looked like a Maya site, and Mexico’s archaeology agency found no entry at the location in its register.
Miotti later corrected the story: archaeologists had already mapped the site in a 2025 paper. He says both he and the agents missed it. A registry check and a search of published research answer different questions; his interpretations of individual structures remain speculative.
The 339-comment thread on Hacker News adds questions about the workflow and its consequences.
What the thread adds
- hanoz, who says they created a UK lidar map, emphasizes the role of growing public lidar coverage: “this story still feels like a function of the amazing recent progress in lidar coverage and data availability.”
- TSiege worries that locating sites faster than archaeologists can protect them could aid looters. timdiggerm counters that looters can use the same process too.
- zingerlio asks about the setup for running 48 agents; joshmarlow asks how they coordinated and shared data. The account does not detail the full setup.
HN handles are pseudonymous, and no per-comment scores are published. This is a slice in HN’s ranking order, not a consensus.
The November 11 afternoon AI News also covered the correction.