Time for amateur birdwatcher qualification?

Wildlife photographers' "obsession" with image quality could be detrimental to birding research, warn citizen scientists 

By Hillary K. Grigonis

The drive for better bird photos is actually contaminating valuable resources, researchers warn in the new era of AI photo editing. As a photographer I'm always striving to level up my images, but researchers are warning bird photographers that one way to "improve" a bird photo may actually end up contaminating valuable citizen scientist databases with misleading data. In a recent op-ed from the journal Nature, researchers from citizen science databases including iNaturalist and the Macaulay Library told a startling story of a Red-Winged Blackbird found well outside its usual range in Brazil. The problem? The bird was actually a common Epaulet Oriole but, when the photographer asked an AI platform to "improve this photo", AI removed the bird's yellow coloring, making it look like an altogether different bird. While fake wildlife images are already littering social media platforms, the altered images' existence inside these so-called citizen scientist platforms could have serious implications, as researchers will use the sightings reported from hobbyist birders to monitor a species' habitat and patterns. The image that the researchers used as an example wasn’t from an internet trickster, but rather from a birder who didn’t realize the implications of asking an AI platform to "improve this photo".  GenAI platforms often don't edit a photo in the traditional sense, like improving contrast, exposure and color. Instead, some GenAI photo editors regenerate, turning a real photograph into a generated image. In the world of birding, the AI can introduce subtle 'enhancements' that actually change the identification of the bird completely.  "AI image processing tools are now ubiquitous, and there is a pressing need to educate contributors of the problems that image manipulation can cause for the scientific record," the researchers wrote. A growing number of photo editing apps can now use AI to help novices edit photos without the usual know-how. But the apps don’t always make it clear when AI is editing and when it is regenerating pixels – and many don’t come with an option to turn the AI features off. The researchers call for the citizen scientist platforms to integrate AI detection algorithms – some of which have already been implemented, such as recent changes to iNaturalist. But the large number of different GenAI platforms can make detection challenging. For birders, the message is clear: don't use GenAI to share wildlife sightings on community platforms for sharing sightings.