Large language models (LLMs) offer a transformative solution to one of the most persistent challenges facing natural history collections: converting textual locality data from specimen labels into precise geographical coordinates. Traditional georeferencing methods require substantial expertise, time, and financial resources, constraints that have left millions of specimens in museums and herbaria without spatial data. Our standardized testing demonstrates that contemporary LLMs can achieve near-human accuracy in georeferencing tasks while dramatically reducing both processing time and costs. By integrating LLMs into existing digitization workflows, institutions can accelerate the spatial enablement of their collections, unlocking new research opportunities in biodiversity science, climate change studies, and conservation planning. This approach has immediate practical applications for collection managers while advancing the broader goal of making natural history data more accessible and analytically powerful for the research community.

TAKEAWAYS
Performance benchmarks and validation
Understand how LLM accuracy compares to traditional georeferencing methods across different specimen types, locality descriptions, and geographic regions, with concrete metrics from standardized testing.
Scalability and resource optimization
Discover how LLMs can dramatically reduce georeferencing bottlenecks, enabling collections to process thousands of specimens at a fraction of the traditional cost and time investment.
Unlocking collection value
See how accelerated georeferencing opens new avenues for research applications, data mobilization initiatives, and cross-institutional data sharing that were previously limited by spatial data gaps.

Register for the Event