What's the best time to prune roses? What are good companion plants for tomatoes? How do you plant a fruit tree?
When we have questions about gardening, it's second nature for many of us to turn to a search engine like Google. These days, that means your question might be answered first by Google Gemini, an AI chatbot that packages the search results into a neat 'summary.' AI tools like Gemini, ChatGPT, and others can help identify plants, offer care advice, and even generate plant lists and comprehensive garden designs. The real question is whether this information is trustworthy and reliable.
As artificial intelligence makes its way into more and more areas of our lives, it's increasingly important to be critical of the information these tools provide. As social media platforms burst with more and more AI-generated plants, we sometimes get asked for plants that don't actually exist, like bright purple hostas, blue sunflowers, or the 'Moonlight Butterfly' begonia. At Cornell Farm, we've even encountered guests who have used tools like ChatGPT to design their entire gardens — to mixed degrees of success. As this new technology becomes more and more common in our everyday lives, join us for some food for thought as we explore the role of artificial intelligence in the garden.
How AI Works
When people talk about "AI," they're usually referring to a Large Language Model (LLM). These are complex computer programs that use massive datasets to "learn" how to emulate human speech patterns. Google's C4 dataset contains the text of over 15 million websites, including all of Wikipedia, numerous news platforms, and online libraries containing millions of full-text books (as well as less credible sources like personal blogs and social media comments).
By pulling data from across the internet, LLMs generate answers based largely on the statistical likelihood of those words appearing together in its training data. It's not necessarily consulting the best sources — often, it's grabbing the text that's easiest to find and most frequently repeated. It's like asking a hundred strangers on the street a question and doing the math to generate the average answer, even though few of those people are experts on the subject. AI will often give definitive answers when, in reality, the answer is nuanced. Sometimes, the information will be perfectly accurate; other times, it's riddled with misinformation and inconsistencies. When you ask for the source of the information, LLMs will often be unable to provide the citation — or even "hallucinate" non-existent news articles and academic sources.
In addition to large language models, generative AI like Midjourney and Adobe Firefly produce detailed images, while tools like Google's Veo 3 even generates convincingly photorealistic videos. While fans often focus on the potential of this technology, the ability to fabricate life-like imagery and authoritative-but-unchecked information makes it easier than ever for misinformation to spread like wildfire.
The Good, The Bad, and The Ugly of AI
There's no doubt that artificial intelligence opens the doors to incredible uses, from cancer detection to wildfire mapping. This technology can sift through enormous datasets to detect patterns that humans cannot. Many people report that AI tools help organize their schedules and channel their creativity, freeing up precious time to spend on the most meaningful parts of our lives. In the world of gardening and agriculture, AI-driven technology is already being used to automate everything from watering to fertilizing.
On the other hand, the rapid rise of this technology means it's entering every arena of our lives with few safeguards. One of the major hazards is misinformation. This can be as inconvenient as a garden plan that incorrectly lists a shade-loving plant as full-sun... or as mind-bending as photos that casually change our perception of reality, making it harder to find reliable info. In a report by Google and Duke University, researchers found that AI-generated images have become a growing source of visual misinformation since early 2023 — over 30% of online myths and hoaxes, going far beyond the world of plants.
Another concerning trend is the fact that the data used to train AI is often harvested from authors, photographers, and artists without consent. The information that comes through AI may sometimes be useful, but it often replaces the original sources of information with no credit — plus, it's flattened of the specificity and nuance that human authors keep in mind as they're writing about their areas of expertise. Large language models scrape content from websites like CornellFarms.com, where we strive to publish quality, place-based advice specifically for gardeners in Northwest Oregon. The result tries to be universal at the cost of being misleading or even flat out inaccurate.
Another unfortunate side effect of AI imagery is how it makes us question our natural sense of wonder. Plants are incredibly diverse, beautiful organisms that come in an astounding range of forms, colors, and patterns. Horticulturalists are constantly breeding new varieties to test the limits of what's possible — for example, the bioluminsecent 'Firefly' petunia is a real plant that has been genetically engineered to glow in the dark. One of our favorite annuals this year is the otherworldly 'Celebrations' kangaroo paw, a cultivar which almost looks as if it evolved in an alien ecosystem (pictured above). Many of today's most popular garden plants have been lovingly bred to coax out brilliant colors and unique behaviors, from electrifying coleus patterns to densely-petalled reblooming roses. Even without the artificial allure of AI, gardening is an art full of astounding beauty.
For eco-conscious gardeners, one of the biggest issues with AI is the hidden environmental impacts. AI uses an enormous amount of energy — a ChatGPT query is five times as energy-intensive as a web search, though search engines like Google now automatically embed AI-generated summaries in many of their searches. By 2026, North American data centers for AI and cloud computing could use upwards of 1,050 terawatts of electricity and up to 6.6 billion cubic meters of water to cool the infrastructure. If these data centers were a country, it would be the fifth largest consumer of electricity and 61st largest consumer of water in the entire world. Recent research shows these data centers also cause air and water pollution, which create invisible costs for local communities and the environment. The environmental impacts are an important consideration as AI grows because, even though we interact with ChatGPT and Google Gemini online, this technology has a physical footprint in the real world.
Staying Wise to AI
Whether you use artificial intelligence to aid your gardening is a personal choice. There are benefits and downsides to weigh, but it's important to be aware of how this technology works before we rely on it to answer our most pressing questions. It can be tantalizing to save yourself the effort of designing an entire garden by hand, yet it's worth knowing that AI has real limitations. The information is not always accurate or relevant. While you could potentially save time and money by using AI for garden advice, you may also waste your resources on plants that won't actually thrive in your conditions.
It can be hard to spot misinformation on topics you're just learning about, so you want to turn to the most trustworthy sources possible and double-check any information you have questions about. Even without AI, there are countless myths and outdated pieces of advice in the world of gardening. Gardening is somewhere between an art and a science — some questions have definitive answers, while others benefit from hands-on experimentation.
At Cornell Farm, many of our staff have strong backgrounds in horticulture and we are constantly learning more about plants. Between hands-on experience and helping local gardeners trouble-shoot their plant questions each day, our team is armed with specific knowledge for gardening in Portland Metro area. On our website, we create quality resources and draw from as many high-quality sources as possible to give you grounded, up-to-date info on best practices for reaching your garden goals. We curate books on everything from garden design to technical how-tos. One of our favorites is the Tilth Alliance's Maritime Northwest Garden Guide, a wonderful resource for beginners and experts alike which walks you through a full year of Pacific Northwest gardening.
In general, university extension services and regional governments around the country are great sources for quality plant facts. Oregon State University's Extension Service is one of our favorite resources for garden guidance that's both local and backed by science. OSU's research teams are constantly developing high-quality resources for both agriculture and home gardening. Their website is well-worth exploring and their certified Master Gardeners are living treasure troves of information.
Gardening is always a process of discovery. While AI may be able to help you learn and plan, there's no replacement for getting your hands in the dirt, observing how different plants grow in your unique space, and connecting your real-life human friends and neighbors — plus, the friendly experts at your local garden center. In this rapidly developing world of technology, we always encourage you to be critical, stay grounded, and keep growing.
Additional Sources
Google and Duke University's Research on Misinformation and AI: https://arxiv.org/pdf/2405.11697
Where Google's AI Data Comes From: https://www.washingtonpost.com/technology/interactive/2023/ai-chatbot-learning/
Article on AI Video Generation and Misinformation: https://www.aljazeera.com/economy/2025/6/26/googles-ai-video-tool-amplifies-fears-of-an-increase-in-misinformation
Tips for Identifying AI-Generated Images: https://www.npr.org/2023/06/07/1180768459/how-to-identify-ai-generated-deepfake-images
(Header image created through Canva's generative AI)
