AI chatbots produce sanitized rewilded Britain images
Get the Health newsletter
Daily health & science — research, biotech, public health, the studies worth knowing. Free.
- University of Aberdeen geographers generated rewilded UK landscape images using AI chatbots (Gemini, ChatGPT, etc.) after prompting with general requests such as “what rewilding in Scotland looks like”.
- AI chatbots produced images dominated by distant hills, open meadows, streams, golden light, and iconic animals (ponies, deer, Highland cows) while omitting humans, buildings, predators, and decaying matter.
- The researchers found that only highly specific prompts detailing species, habitats, and landscape features could coax the models into rendering ecologically accurate, messy scenes with wolves, lynx, scrub, and decay.
- AI models draw visual inspiration from publicly available sources, including the social media and websites of environmental NGOs such as Cairngorm Connect and Knepp Estate Rewilding, which often showcase sanitized, aerial perspectives.
- The article warns that these tidy, human‑free depictions could set public expectations of neat nature recovery, making the inevitable scrubby, predator‑rich reality of rewilding harder for communities to accept.
- Historical context is invoked, noting past shifts from Defoe’s disdain for wild landscapes to Romantic sublime imagery, and highlighting that AI’s “sublime” viewpoints still lack the ugly, untidy aspects of real rewilded sites.
Why it matters: If AI‑generated images set a public expectation of neat, picturesque rewilded land, then real‑world projects that produce scrubby, predator‑rich habitats may face aesthetic backlash, undermining community acceptance and funding for conservation initiatives. This dynamic could benefit NGOs that already market sanitized visions, while disadvantaging researchers and land managers who must contend with ecological realities and public skepticism.
