ChatGPT for Text
When prompted by “How to water a garden in street slang” the result included “H20 to hydrate the greenery, fam.”
“How to plant a tree in the style of hiphop” yielded standard tree-planting instructions but also “Give the tree some love and positive vibes, just like a rap lyric” and “Represent the tree with rhymes and lyrics, spreading its message.” Pretty lame.
My prompts for “How to prune hydrangea” and “How to design a townhouse garden” produced results we see everywhere – generic, not wrong. Too boring to include here.
I did enjoy the result for “How to water houseplants in the style of Shakespeare,” which used “thy” repeatedly – “Fill they watering can” – but added some philosophical tips I enjoyed: “Tend to them with love, for they bring beauty to thy home with ease” and “Thus shall thy plants prosper and grow, fair and strong like the trees.”
And here’s “How to water houseplants in the style of Melville”:
- With a steady hand, grasp thy watering vessel.
- Pour a measured amount, as if to nourish the earth’s treasure.
- Let the water flow deep, penetrating the soil with leisure.
- Give time for the excess to depart, lest it remain and spoil.
- Be mindful of the frequency, for over-watering is a dangerous foil.
- Observe the leaves, if they droop, then water with all thy power.
- Provide them light and air, for these are their daily dower.
- Tend to them with care, for they bring a touch of nature’s bower.
- And thus, with patience and love, thy houseplants shall flourish hour by hour.
Everyone is testing Chat GPT, including the very literate, hard-to-please hosts of Slate Culture Gabfest. In this episode, they were surprisingly positive in their review of the app, calling it “capable of writing a decently cogent college-style essay.” One host prompted it to write a five-paragraph essay about a particular Wordsworth poem and declared that the result would deserve a B+ or A- in a high school or freshman class in college class. Another prompt – to compare two philosophers – produced “a quality paper for a sophomore at a top college.” This from a guy who’s taught English at top colleges.
Yikes! No wonder there are signs of panic over the app in academia.
On the downside, Slate’s film critic reported absurdly inaccurate statements about film history that produced by the app.
The podcasters concluded with praise for the app’s press-release-writing skills. I tried that myself and hell, yeah, if I ever have to write another press release, I’m starting with Chat GPT’s version and tweaking it from there.
So will we see garden writing by Chat GPT? Oh, you bet. Content mills are already publishing articles generic and inaccurate enough for us to wonder if they were produced by artificial intelligence, and Chat GPT will save them money.
DALL-E for Photos
“DALL·E‘s website proclaims that it “can create original, realistic images and art from a text description. It can combine concepts, attributes, and styles.” And that may be underselling its abilities.
Here are some images it created in response to various gardening prompts. Clockwise from upper left they were “koi in pond,” “lawn with clover and pony,” “black cat in garden with roses,” and “lantana in pot.” For each prompt the app creates four options, which can be altered within the program.
As a cat owner who sadly can’t show off cool photos of them in the garden because they’re indoors-only, this image of a cat that looks just like one of mine is giving me ideas.
So I prompted the app with more details about my cat family, using “a short-haired black cat and a long-haired black cat in a garden,” and these results are pretty darn close.
Bottom line, I’d assert that (almost) any of them would be preferable to having NO image to illustrate a gardening blog post. But feel free to disagree.
Wanna play with styles? Here are four responses to the prompt of “succulent plants in the style of Monet.”
And “rain garden line drawing.”
Just for fun I asked DALL-E to illustrate something that doesn’t exist – cactus growing in a meadow – and the results may tempt even a normally honest gardenblogger to try to fool her readers. So look out!
It certainly doesn’t know Cockney slang, or indeed, Yorkshire dialect, having tested that. But those things may come.
What I wonder, with some tiny feeling of hope, is that it might kill off forever the bland anodyne cliché ridden garden writing that fills all our garden media. What editor would pay – even the usual pittance – for some of that when they can get it free?
Or maybe they’d even consider breaking the mould and demanding interesting, original garden writing???
As if…….
A few weeks ago, I asked it to write an article about five American women garden writers born before 1900 and got back a list that included a man, two British women, and a writer born after 1900. One of the books credited to one of the authors was clearly not written by her. Then I asked it to write an article about the podcast I record weekly… with lots of praise. Interesting but it used the wrong names for me and my co-host. Twice. Different names each time. So not too accurate!
With the latest garden catalog, my heart raced as I fondled the pages. Caressing the spine, excitement surged with each new picture of voluptuous blooms, fantasizing about a future summer night where the flowers produced a heady scent.
(In the style of E.L.James, 50 Shades of greenery?)
Sorry but I find this use of technology pretty sad. If I am going to read something I would prefer it came from a knowledgeable, insightful and informed human. Same with the photos. A picture speaks a thousand words so if we can create any content we desire would make it even harder to try and duplicate. Perhaps a similarity to the fashion industry? Hmm
Mixed feelings. As for the errors, ‘garbage in, garbage out’. I am sure it will continue to be perfected but I see it possibly as a reference tool as I construct my own columns. I don’t believe any AI will truly replicate my style or anyone else’s for that matter, at least not until they can program in a healthy degree of snark and irreverence. I do not feel threatened in the least.
If I were inclined to play with AI, I’d ask it to construct an airtight argument as to why AI, used as a tool for original content, is a terrible idea and, given its findings, to do the honorable thing.
Belly laugh from Marianne. 🙂
I heartily endorse Joe Schmitt’s idea.
I have seen cactus growing in, if not meadow, at least pasture. And a neighbor of my late parents, in Beaverton, OR, used to have rather a lot of prickly-pear – 10 or a dozen plants – in their front garden-bed. Yes, it got rained on; yes, it occasionally went moldy. But cactus in a meadow isn’t an impossibility!
Hmmm, as a garden content creator, I suppose I should feel threatened, but until AI can come and give live garden club talks and book signings, I think I’m safe. 😉
I can see the photo AI being of some use to say illustrate a bed or combination that you want to design “show me tulips with hellebores.” As far as the text AI, as you say writing a first draft of something like a press release or a podcast episode recap that you then use to edit, could be a real time saver.
Make technology work for you.
A fascinating post – thank you Susan! I was actually not aware of these tools but have been experimenting today. This feels like legitimately disruptive technology that has the potential to impact a broad swath of sectors, especially now that these programs are easily accessible and super easy to use. Minimally, I can see why academia is freaking out about this. I’m equal parts fascinated and unsettled about this technology, and will be following its progress closely.
Opuntia in scrubland reminds me of the Waikoloa area on the Big Island of Hawaii.
The 4th picture captures it perfectly.
I’m going to chime in on the cactus in a meadow… we have that here in Michigan on our dunes. Prickly pear cactus grows in the grasses on our dunes and makes for interesting walks if you stray from the path!
Susan and GR fans—when you have a moment you may enjoy reading the following:
https://engineering.jhu.edu/news/where-is-chatgpt-taking-us-and-do-we-want-to-follow/
Great post, Susan! I was unaware of DALL-E.
I’m afraid Garden Blogging is doomed however. Not that I think anyone reads mine anyway.
Great post Susan – thanks for digging deeply on this. I’ve got lots of thoughts but there’s a lot of profanity involved, so I’d better wait and formulate something a little more thoughtful. – MW
Oh wait, I’ve just had a good thought. If garden blogging is doomed because we can start to assume that anything we read is AI generated, perhaps it means a resurgence in books, and the ultimate gatekeeper – the publisher? There may be a silver lining after all…. – MW