Brazil’s Geological Survey Launches GeoBusca, a Generative-AI Search Over 5 Million Research Records
The Geological Survey of Brazil (SGB) has launched GeoBusca, a search platform that uses generative artificial intelligence to help people find studies and products drawn from about 5 million records built up over more than 57 years of research, the agency said on November 2. The tool was presented at the 52nd Brazilian Geology Congress and is available as a beta, a phase the SGB says is meant to evaluate how it performs and improve it.
Instead of relying only on keywords and jumping between several websites, users can ask questions in everyday language and receive suggestions of relevant material with a short summary. Each result lists its sources and sends users to the survey’s official systems, such as the institutional geosciences repository RiGeo, the critical events alert system SACE and GeoSGB. The SGB stresses that people should open the full documents in those databases to confirm the information. The collection spans basic geology and mapping, mineral research and potential, including critical minerals, surface and groundwater, river monitoring, geological risk areas and territorial management, along with geophysical surveys and geochemical data.
“Geoscientific knowledge reaches further when it can be found, used and applied,” said SGB chief executive Vilmar Simões, adding that decades of research can now be located faster and more simply. The platform was built by the agency’s communications office.
What it means in practice
Engineering news site Giro Engenharia framed the launch around critical minerals such as lithium and rare earths, where quicker access to historical surveys could speed up early exploration work by companies and researchers. The tool joins other public efforts to put AI to work on national data, from the UFMG Soberano supercomputer built to map Brazil with AI to research hubs such as the IPT’s R$20 million MaterI.A. centre for advanced materials.
Generative search can produce confident but wrong summaries, which is why the beta label and the links back to source documents matter. What to watch is how the SGB measures the accuracy of answers, whether it offers programmatic access for companies and universities, and when the platform leaves beta.