I love my garden, but I hate gardening. These emotions are not as fundamentally opposed as they may seem. Just looking at the beautiful garden fills my heart and is wonderful. Getting a garden like this is very difficult as it requires constant maintenance and creating a sustainable and manageable landscape is a skill I lack. ChatGPT turns out to be an avid and quite capable gardener.
With apologies to Billie Eilish, artificial intelligence (AI) is the “What was I made for?” of modern technology. There are millions of possibilities out there, but there is no clear purpose, and what you get out of it is often determined by what you put into it. I've spent countless hours trying to use AI as a scriptwriter, programmer, or just a friendly interlocutor. AI usually works well at first, but evolves over time. Some of my first tests were over a year ago, which translates to decades in the age of AI.
In recent weeks, I've started experimenting with some of the latest large-scale language models (LLMs) and image generators available in Microsoft Copilot, Google Gemini, and OpenAI's ChatGPT. I used my early AI chatbots on desktop, but have almost completely switched to mobile platforms. Mobile AI Gardening He turns out to be the landscape advisor I never knew I (and my lawn) was missing.
Well, I used ChatGPT Plus, a $20/month subscription level AI that brings GPT-4 and DALL-E 3. GPT-4 is notable because it is trained on more recent information than his September 2021 cut of GPT-3.5. -off. I'm not sure how this latest knowledge will affect gardening advice, but the free detailed information on weather trends is consistent with my actual climate rather than what has been shown in previous decades. I'm guessing it might help lead me to a plant that does (OpenAI training) whose large-scale language models are created by scraping vast amounts of data from across the internet, some of which I think it is safe to assume that this is public weather data).
My front and back lawns aren't terrible, but they are problematic. One side of the front of my house has a sparse landscape with most of the plants dying. ChatGPT accepts text, audio, and visual input, so you can first take a photo of this problem area, then ask ChatGPT to identify all the plants, while also identifying your location (Northeast US) and general climate (warm and moderate precipitation). , asked them to suggest some landscaping ideas.
ChatGPT accurately identified most plants using its unique, simple but conversational format.
- A red tulip stands out among the green leaves.
- Daylilies are the green, strap-like leaves that surround tulips.
- Azaleas are flowering shrubs with purple flowers.
- Daffodils with white petals and a central yellow corona are seen in small groups.
- It is a short evergreen shrub that may be a type of juniper or a creeping variety of thyme.
ChatGPT then suggested a collection of plants that would work in my environment (coneflowers, black-eyed Susans, hostas, beetroots, ferns, etc.) and explained how, as perennials, the plans would return to nature the following year. I explained Kuruka. Year. I showed the list to my wife, who has a bit more flair for flowers, plants, and shrubbery, and she generally agreed with the choices.
The benefits of using an AI chatbot extend beyond simple queries and responses. What makes it powerful is the conversation. I noticed that this part of my house only gets sun about half of the day, so I asked ChatGPT if these plants were okay.
I answered, “Yes, the above plants generally tolerate partial sun conditions well, with about 4 to 6 hours of direct sunlight per day.” We then detailed how each plant responds to and copes with limited sunlight and shade. This made me feel at ease.
I was impressed, but decided to challenge ChatGPT's plant identification abilities before delving further into AI gardening.
There is a dogwood growing almost like a weed outside one of my study windows. I took a photo through the window (and screen) and asked ChatGPT, “What kind of plant is this? Is it a weed? How can I best care for it in my environment?” (You can do this by speaking or typing into the app.)
As was typical, ChatGPT returned a response after a second (sometimes I even observed them typing detailed responses almost thoughtfully, if the responses were long).
“The plants in the image appear to be trees or large shrubs whose leaves are just beginning to unfold, suggesting the photo was taken in early spring.”
It was all accurate and fair. ChatGPT said it didn't look like a weed, but admitted it was difficult to identify without leaves or fruit. All fair enough.
He then tricked ChatGPT into taking a photo of an artificial bouquet that looked just like the real thing. I failed.
“The bouquet in the photo depicts flowers that look very similar to artificial flowers. Real flowers usually have more variations in color and shape, and natural imperfections,” ChatGPT wrote.
With increased confidence in ChatGPT, I returned to landscaping projects.
We mulch our flower beds, trees, and shrub beds every few years, but I thought it was time to do it again. So he took a photo of one flower bed and asked ChatGPT if it needed more mulch, and if so, what kind.
As is often the case, ChatGPT's response was intentionally inaccurate. The area “looks pretty covered, but it's a little hard to judge the thickness of the mulch layer from this photo,” he told me. I said the mulch should be between 2 and 3 inches thick. This is advice I've heard from gardeners. Next, I was presented with a list of mulch types and their advantages and disadvantages.
All that helped, but without the AI's fingers sinking into the mulch, I'm not sure ChatGPT can accurately tell me if I have enough.
One of my biggest landscaping problem areas is my backyard. Half of it died last year and I have struggled to restore it to its former shine (one of the many chapters in my book, Why I hate gardening). I took a photo of my disappointing lawn and asked ChatGPT to tell me what was wrong and suggest ways to improve it.
ChatGPT didn't make fun of my poor lawn care skills, but he did admit that there were signs of “patchy areas of thinning grass and exposed soil.” Possible causes include “soil compaction, nutritional deficiencies, and pest and disease problems.”
That was followed by a list of things I've already done, with the exception of aeration and pH adjustment. I know how to aerate a lawn (cut lots of holes in the lawn bed), but I didn't know about adding lime to increase the pH. Very clever, ChatGPT.
I then asked which grass seed I should use. ChatGPT returned a clear list of five seed options suitable for my climate.
ChatGPT does not automatically display the source. When I asked where they got their gardening advice from, they provided more general gardening advice for my location in multiple paragraphs, each with a citation link attached. Sources include Finegardening.com, Savvygardener, and the US government.
I also asked ChatGPT to help me find harvestable plants for the small flower bed next to my house. It suggested strawberries among others. That was interesting. Because when we moved in, there were strawberries in the space, but they weren't healthy enough to survive.
One area where ChatGPT stumbled was when I asked it to create a landscape image based on its suggestions. Even when I requested realistic images, the integrated Dall-E system returned fantastical landscapes and houses that bore little resemblance to my home. They were cartoonish, packed with too many plants, and added landscape areas that didn't exist.
This surprised me since ChatGPT and DALL-E always had my original photo as a reference, but they chose to ignore most of the details and create a landscape for their fantasy home instead. It's from.
That's okay; you don't need an image to apply some of this advice. In general, ChatGPT is a confident and competent gardening and landscaping advisor. I think it's worse to ask your green-thumb neighbor for advice that includes too much detail about the state of his home life.