Answers are grounded in this paper and cite their sources.
Entomologists generally agree that we need more biodiversity data if we hope to address global concerns like climate change, mass extinctions, and food security—and yet it seems like we’re doing less of that kind of research than we used to. We all know why it’s happening: it’s the “taxonomic bottleneck” or “taxonomic impediment,” whatever we want to call it (Kim and Byrne 2006, Engel et al. 2021). Identifying arthropods is no trivial matter, and the experts who’ve mastered the craft are becoming increasingly rare (Kim and Byrne 2006, Orr et al. 2020). So, in this time when biodiversity data is more important than ever, we’re less well situated to collect it than we’ve ever been before. We need a way to fill that gap. One potential solution to this problem is artificial intelligence (AI). Image-recognition models that can rapidly classify objects from photographs into categories are readily available. In theory, we should be able to apply this to the important work of studying arthropod diversity. Entomologists have been experimenting with this technology for decades (e.g., Weeks et al. 1999, Mayo and Watson 2007, Fujisawa et al. 2023), and public-facing “bug ID” apps are becoming common. Many of these apps are amazing resources for amateur naturalists. However, if we’re being brutally honest, we haven’t yet converted this technology into a tool that makes the research process more efficient. That’s something that needs to be done.
No takes yet. Share an insight, caveat, or question.
Welch et al. (2024) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: