When AI misidentifies wildlife: New study maps where species identification fails
BiologyEnvironmental Science
THE AI ANGLE
Classifying animal species in wildlife photographsA test of an artificial intelligence model on 10,000 Colombian vertebrate images yielded a species identification accuracy of under 10%. Error rates were highest for amphibians, birds, fish, and recently described species. Biologists must combine automated image screening with known ranges and expert reviews.
THE TEACHING ANGLE
Forcing automated models to classify every image without an option to abstain risks inserting false geographic data into conservation records.Read the original at phys.org Generate teaching or study materials
More in Biology
- How brine processing could offset sulfur hexafluoride treatment costsPhys.org — Technology · November 18, 2026
- AI tool counts mosquito eggs in smartphone photos to aid surveillancePhys.org — Biology · November 17, 2026
- From 400 miles above Earth, AI tracks invasive leafy spurgePhys.org — Biology · November 17, 2026
- AI and experiments reveal impurities hindering waste plastic pyrolysis oil recyclingPhys.org — Chemistry · November 17, 2026
- Data centers may consume far more water than developers suggestPhys.org — Technology · November 17, 2026