Skip to main content
Entomology 2025 Main banner
Icon Legend
This session is not in your schedule.
This session is in your schedule. Click again to remove it.
 
Oral Presentation
Oral Presentation
Poster Display
Poster Display
Student Competition
Student Competition
Ticketed Event
Ticketed Event

10-Minute Presentation

Systematics, Evolution, and Biodiversity

10-min: SysEB, Invasive Species, Diagnostics, and Pest Control

BugBox: Lessons learned from four years of AI-based arthropod biodiversity monitoring

Sunday, November 9, 2025
8:36 AM - 8:48 AM Pacific
Location: Oregon Convention Center, E147-148, OCC
Oral Presentation
Over the last four years, Ecdysis Foundation has been conducting our 1000 Farms Initiative, which is an extensive, multi-faceted evaluation of regenerative agricultural practices in various food-production systems across the United States and Canada. As one facet of this research, we have collected and photographed over 650,000 arthropod specimens. We developed an artificial intelligence (AI) platform called BugBox to identify these arthropods to the morphospecies level. We have found that the AI can provide useful biodiversity assessments, in the form of calculated diversity indices, which have allowed us to test hypotheses about the beneficial effects of regenerative agriculture on arthropod biodiversity. We have also found that, even while the AI’s mastery of the subject matter is low (i.e., it only knows ~700 morphospecies of arthropods, and identifies them with ~75% accuracy), it still uncovers the same trends in the data as a human identifier. Here we discuss our ongoing review process, in which human experts evaluate and correct the AI’s identifications to improve its performance over time. We also discuss the development of a familiarity index to account for the AI’s knowledge gaps, and we report some findings related to AI training protocols that have worked for us. We also report new developments in multi-specimen imaging and integration of specimens from external sources. BugBox is now available for other scientists to use, and it is an excellent way to introduce new students to biodiversity analyses and artificial intelligence. We invite other scientists to begin using BugBox for their own projects.