Gardening Meets AI

Artificial intelligence (AI) is being widely used and has even found its way into gardening.

An AI-generated image of multicolored hibiscus flowers. — An AI-generated image of hibiscus flowers. Take note of the unnatural colors and the overly blurred background, clues to the image’s inauthenticity.

What was once science fiction has now become a part of our daily lives. From innovative “smart garden” networks, plant identification apps, and AI-generated botanical images, there’s a lot to sort through in the modern world.

AI-Generated Plants

Have you recently come across any out-of-this-world plant images that seem almost too good to be true? You are probably scrolling through AI-generated content. Generative AI uses machine learning or computer programs to generate abstract or realistic content. It pulls users’ online content to generate images, audio, and video. While this content can be used for increasing efficiency, inspiring creativity, or enhancing communication, the misuse of generative AI can lead to scams, fraud, and misinformation.

The difficulty in identifying which images are AI-generated and which ones are real can be tricky business. While there are some clues to look out for, it’s not always obvious, leading people astray. Individuals are creating images of outlandishly colorful, oddly shaped, or oversized plants, often to peddle seeds for these nonexistent plants. These people are taking advantage of the human desire for beautiful and interesting plants. Nursery owners have even reported customers coming in requesting a plant that does not exist.

An AI-generated image of a yellow rose-like flower. It's difficult to tell if the lowest layer is made up of petals or leaves. — Are those leaves or petals at the bottom? Another AI-generated “flower.”

To avoid being scammed, look for certain details in the photo that might point to it being generated using AI. Many AI-generated photos are close-up shots with blurred backgrounds, which is done to prevent errors in the image. The plants can also appear too perfect, whereas in nature there are often subtle differences within patterns and shapes. Look out for unnatural details, which could include flowers without stems or leaves morphing into the flower. The plants can also appear much larger than they do in nature, so be wary of larger-than-life plants.

Before purchasing a plant or seeds online, make sure they’re from a reputable vendor. If you cannot find the plant listed by a trusted vendor, then it’s likely not real. Plant listings should also provide the botanical name. You can then search for photos of the species or cultivar and compare to the listing’s photos to see if they match up. Referring to reputable plant identification directories or UF/IFAS sources can also help determine the legitimacy of the photos.

AI-Generated Gardening Information

To find answers to pertinent gardening questions you can now ask an AI interface such as Chat GPT or Google’s Gemini instead of searching through articles or books. But what are the pitfalls of this new service? These programs are searching through a host of different sources without distinguishing between commercial-driven content versus scientific information. The response that it formulates may not necessarily be from qualified organizations or experts and can be inaccurate and misleading. You can use the phrase “…using UF/IFAS Extension resources” when typing in your question, which often prompts the AI interface to use information from UF/IFAS documents. But while the AI responses can help you narrow down options, they should still verified. Luckily, a UF/IFAS Extension agent can personally address your unique situation.

Global Implications of AI in Science Communication

AI-generated images and videos pose many opportunities and risks associated with the field of ecology and environmental science. While AI has the potential to improve communication with the public and enhance environmental research, there are implications with fabricated image-based data (like photos purported to be from a microscope) and the spread of fake facts about the environment.

For example, there was an incident in 2023 where books on foraging were being sold on Amazon that were written by chatbots. These books made dangerous and inaccurate claims, such as suggesting that foragers could taste-test wild mushrooms to identify the species. This is a matter of life and death, with toxic and sometimes lethal mushrooms growing in the wild. As with most new technology, AI in the wrong hands can have drastic consequences, so it’s important to approach the tool and its outputs with caution.

Apps Using AI to Identify Plants

In recent years, smart phone applications (apps) have been developed for plant identification using AI to assess photos submitted by users. These apps help determine the genus or species with certain levels of confidence. But how accurate are these apps at correctly identifying the plant? Fortunately, studies have been conducted to uncover this.

Screen shot of plant ID app on mobile phone showing a yellow flower. — Smart phone apps like iNaturalist can be helpful.

One study’s research published in 2023 assessed five popular and free plant identification apps. Researchers used PlantNet, LeafSnap, PlantSnap, iNaturalist, and Google Lens to identify 277 species. Both PlantNet and LeafSnap outperformed the others and accurately identified the plant on their first choice roughly 87% of the time. And while iNaturalist had a slightly lower identification accuracy overall, it was able to correctly confirm many of the images with high confidence. This means when a plant identification in the app reaches a certain confidence threshold, the app will label their answer as “confirmed.” iNaturalist confirmed 66% of the images, and they were correct in all cases with no false positives.

Results of this study and others like it have suggested that plant identification apps are relatively reliable tools for identifying plants, and with the aid of a good guidebook users can confirm the options provided by the apps. These apps can serve as a jumping off point for any gardener looking to identify that mystery plant that popped up in the garden.

(Note: By discussing specific plant apps, UF is not endorsing those apps, just reporting on the research findings of ID accuracy.)

The space where AI and plants merge can be an immense opportunity for knowledge and research, such as the use of AI to collect and analyze data on crops. But stay aware and look out for the dark side of AI — use the resources available from UF/IFAS publications and your local County Extension office to confirm horticulture facts and information.

Also on Gardening Solutions

References

Enroth, C. & Illinois Extension. (2024, February 9). Exploring A.I.’s Role in Gardening Advice. Good Growing. Retrieved January 9, 2025.
Hart, A. G., Bosley, H., Hooper, C., Perry, J., Sellors-Moore, J., Moore, O., & Goodenough, A. E. (2023). Assessing the accuracy of free automated plant identification applications. People and Nature, 5, 929–937. https://doi.org/10.1002/pan3.10460
Milmo, D. (2023, September 2). Mushroom pickers urged to avoid foraging books on Amazon that appear to be written by AI. The Guardian.
Rillig, M.C., Mansour, I., Hempel, S., Bi, M., König-Ries, B. & Kasirzadeh, A. (2024) How widespread use of generative AI for images and video can affect the environment and the science of ecology. Ecology Letters, 27, e14397.
Available from: https://doi.org/10.1111/ele.14397
Zeyu Lu, Di Huang, Lei Bai, Jingjing Qu, Chengyue Wu, Xihui Liu, and Wanli Ouyang. 2024. Seeing is not always believing: benchmarking human and model perception of AI-generated images. In Proceedings of the 37th International Conference on Neural Information Processing Systems (NIPS ’23). Curran Associates Inc., Red Hook, NY, USA, Article 1105, 25435–25447. https://doi.org/10.1145/3610085