Fighting a new invasive species: AI's impossible flora

By Giovanni Aloi / edited by Rachael Petersen

Ten-foot-tall hostas and bright teal blooming geraniums, purple sunflowers, blue pampas grass plumes, black marigolds, and rainbow roses—the list is endless. Social media platforms have recently become bourgeoning virtual gardens filled with “impossible,” AI-generated plants in perennial bloom. Plant lovers have long been cautious of overly photoshopped images, but recent advancements in AI have transformed the online botanical landscape—though not always for the better.

Since its inception, the Internet has had both positive and negative effects on our relationship with plants. Once difficult to access, archives of botanical images, facts, stories, and care instructions are now available to everyone at the swipe of a thumb—and often for free. Apps like iSeek, PictureThis, and PlantNet have made plant identification much easier, educating millions about the vegetal world in new and accessible ways. The house plant fever that swept the planet during the pandemic would never have been possible without social media. Until recently, cuttings of variegated Monstera deliciosa were an Internet sensation, exchanging hands for over $1000 a piece. For a few years, the Internet has been a greatly rewarding and exciting botanical resource for all—but matters have rapidly changed.

As with any subject, the reliability of online information is a constant challenge. Who created the content? Who verified it? Who can we trust? For about a decade, traditional institutions—such as botanical gardens like London’s Kew or the New York Botanical Garden—have had the upper hand, publishing peer-reviewed scientific research in accessible formats. 

However, elsewhere in the online commercial metaverse, the situation is more thorny: “fantastical blooms” have abounded on eBay, Etsy, and Amazon for over a decade. The colors and size of these “impossible” plants should instantly scream “fake” to any discerning gardener or even the average plant enthusiast. And yet, if the “sold items” counters—often in the thousands—are a good indication, many people believe, or want to believe, that orange-blooming hydrangeas truly do exist. It is also not a coincidence that these impossible plants are often sold as seeds. 

“Item not as described?”

The seller will have vanished by the time leaves and blooms have revealed their true identity.

But we are all drawn to rare and unusual plants—the remnant of a not-so-distant past shaped by imperialist dreams and colonial nightmares. Delicate and ennobling social status symbols, from Emperor Huizong’s botanical collection[1] to Louis XIV’s Gardens of Versailles, kings, scientists, and even artists have always craved ownership of exclusive plants.[2] While the Internet has made these rarities more readily available to everyone, it has also made it more difficult to discern true rarities from fantastical fiction. 

More recently, to complicate matters further, an AI-generated breed of impossible botanicals has emerged—a new invasive species complete with mock-Latin nomenclature: flawless, garishly colored, utterly soulless. Appearing on a small phone screen, at the swipe of a thumb, these AI-generated plants swiftly pass themselves off as the real thing. Their unstoppable dissemination has significantly eroded any trust gardeners once had in Facebook or Instagram as reliable botanical resources.

Comments on social media posts range from “Beautiful, where can I buy this?” to “The greatness of God is endless.” Many of these comments might be the work of bots, but others are clearly from legitimate human profiles. Amplifying each other, they create an echo chamber of delusion. 

Will the many drawn to the artificial allure of AI-generated plants still recognize and value the subtle charm, natural imperfections, and quiet dignity of the endangered wild species that truly need our protection? Or are AI-generated plants inducing a new kind of plant blindness?[3] Ultimately, these images mislead us in much the same way that heavily photoshopped depictions of female bodies did when they first appeared in the early 2000s. They distort our perceptions, making real living plants seem less remarkable or even undesirable.

Frustrated and dismayed, small armies of plant lovers now systematically post large “FAKE-AI GENERATED IMAGE” banners in the

 comment sections of the social media posts that disseminate these images—and for good reason. Much more than playful flights of fancy, these images portray true algorithmic fleur-du-mal: clicking on the “buy now” links at the bottom of posts unleashes malware on your computer! Despite the numerous complaints, Facebook and Instagram have not taken any action to stop the creators. Given the current alarm over Mark Zuckerberg’s announcement that Meta will no longer enforce fact-checking, the accuracy of online plant imagery may seem like a low priority.[4] However, this issue raises pressing issues with broader cultural and ecological ramifications. Beneath their glossy surfaces, what do these images reveal about our increasing difficulty in distinguishing real plants from AI-generated ones?

First and foremost, they expose how little time most of us actually spend engaging with nature with any kind of attention. These images are splinters of a deeper fracture: our growing estrangement from the natural world, a severance that has kindled climate change and ushered in the sixth mass extinction. Perhaps more importantly, on a smaller scale, they are eroding our collective sense of wonder. It's tragic that every time we encounter a stunning plant online, our first instinct should be to question if it’s real or artificially created. We’ve lost our trust in the beauty before us, and that’s a steep price to pay—especially in a world already overwhelmed by images of loss and violence. This is a future we must prevent from taking root—one we definitely don’t want to see flourish.

Giovanni Aloi, PhD, is an author, educator, and curator specializing in the representation of nature and the environment in art. He is the editor-in-chief of Antennae: The Journal of Nature in Visual Culture and the author of many books on the subject of animals and plants in art. Aloi has contributed to BBC and NPR programs and worked at the Whitechapel Art Gallery and Tate Galleries in London, as well as the Art Institute of Chicago. He currently teaches modern and contemporary art at the School of the Art Institute of Chicago and has curated exhibitions in the U.S. and Europe. He is also co-editor of the University of Minnesota Press series Art after Nature.

[1] Patricia Buckley Ebrey. Emperor Huizong. Harvard University Press, 2014.

[2] Thompson, Ian H. The Sun King’s Garden : Louis XIV, Andre Le Nôtre, and the Creation of the Gardens of Versailles. New York, Bloomsbury, 2006.

[3] Wandersee, James H. and Schussler, Elisabeth E. ‘Preventing Plant Blindness’ The American Biology Teacher, Vol. 61 No. 2, Feb., 1999; pp. 82-86

[4] Schleifer, Theodore, et al. “Meta to End Fact-Checking on Facebook, Instagram ahead of Trump Term: Live Updates.” Nytimes.com, The New York Times, 7 Jan. 2025, www.nytimes.com/live/2025/01/07/business/meta-fact-checking. Accessed 13 Jan 2025.