Has Anyone Used ChatGPT For Help With Bonsai

[leatherback]

AI being used to do a job that a human could (and should?) be being paid to do

Q: Why should we work if technology can work for us?
Sorry, but if the world could operate without people manning desks and I can just do my hobbies and hang out with friends I might actually not object. Or would it not be nice if we could drastically reduce work hours to a few days a week? Naturally this comes with the need for a very different economical model, one where it is not the few haves that own all havenots.

 

[littletreefarmer]

Q: Why should we work if technology can work for us?
Sorry, but if the world could operate without people manning desks and I can just do my hobbies and hang out with friends I might actually not object. Or would it not be nice if we could drastically reduce work hours to a few days a week? Naturally this comes with the need for a very different economical model, one where it is not the few haves that own all havenots.

I agree. I also live in the real world, in the most "free" country in the world.

Do we really, for a single moment, believe that our careers being phased out by this technology will really result in everyone just living for free and doing whatever they want? Has any technological revolution in human history had that impact? The "Haves" have never allowed the economic model to change outside of their favor. I don't buy the equality or democratization arguments, new technologies most often result in more disparity and separation despite what we are told to think. The privileged very seldomly see or care about where the prices of growth are being paid, however.

Regardless, like we both agreed, the new way is now and we either grow and adapt or get left behind, I suppose. Still doesn't make it right, though, but what is right anymore?

 

[JRidge]

Another of the big fears about this technology already being realized.. Do I need to state the obvious that there are indeed humans who are trained in coding? your example is a direct representation of AI being used to do a job that a human could (and should?) be being paid to do.

When will the world need to be so efficient that we as humans no longer have a place? I guess the $$$ means more.

One could of course argue that the aesthetic pursuit of art or beauty does not fit in to the lifestyle and narrative of supreme human efficiency we all seem to be chasing.

Although I actually agree with you… and I do not like the idea of essentially eliminating the human workforce in many sectors, not just coding!
I can’t help but point out that technology has been replacing human jobs for as long as we have existed! The Luddites are a great example of this. But alas we retrained, created new jobs and humanity carries on. Survival of the fittest so to speak.

Luddite - Wikipedia

en.wikipedia.org

Back to this chat tree design…. I interpret it as creating a windswept tree @Bonsai_hope is that what you are envisioning also from the diagram?

 

[stav121]

Q: Why should we work if technology can work for us?
Sorry, but if the world could operate without people manning desks and I can just do my hobbies and hang out with friends I might actually not object. Or would it not be nice if we could drastically reduce work hours to a few days a week? Naturally this comes with the need for a very different economical model, one where it is not the few haves that own all havenots.

I am on your side on this, I would rather do my hobbies and travel than work.

I am kinda disappointed of the current state of technology to be honest. Growing up we were told that by 2020 we would have flying cars and robots doing our jobs for us - where is that? It is 2025 and we can't even get an AI to answer basic questions correctly.

 

[yenling83]

I've primarily used it for help with Pest and Disease Management. For example, I wasn't sure if it makes more sense to use the same pesticides/fungicides for a year straight, then rotate to something new, or if you should rotate every time you spray. Chat GPT showed me the studies which suggest rotating every time you spray is much more effective at not allowing pest/disease to build up tolerance.

 

[Bonsai_hope]

Out of curiosity, what did it tell you to do with this image? Is the dotted lines where you're meant to cut to? What were the arrows supposed to be?

It gave me a detailed list of instructions

Out of curiosity, what did it tell you to do with this image? Is the dotted lines where you're meant to cut to? What were the arrows supposed to be?

This is what it told me before generating the image:

At 16 inches tall, it falls into the Chuhin (medium-sized) bonsai category, which is perfect for a Chinese elm in an informal upright or broom style — both suit its natural growth habit.

Here’s how its proportions look and how you might refine it:

Proportion and Ideal Height

Your trunk looks about ¾ inch thick at the base, so by the bonsai proportion rule (height ≈ 6–12× trunk width), a final height between 9 and 18 inches is ideal.

At 16 inches, you’re already within that balanced range — so the height is appropriate.

You don’t need to shorten it unless you want a more compact or denser look.

Design Suggestions

1. Define a primary line (trunk shape):
   The trunk has a gentle curve — ideal for an informal upright (Moyogi) style. Keep emphasizing that movement by pruning branches that go straight up or cross the interior.
2. Top shaping:
   Consider reducing the top by 1–2 inches next spring to encourage ramification and create a rounded canopy — Chinese elms respond well to trimming and will bud profusely.
3. Branch density:
   Several upper branches could be shortened to 2–3 leaves per shoot to encourage back-budding and denser growth next year. This will fill in the canopy nicely.
4. Lower branch structure:
   Keep a few strong lower branches to balance the visual weight — they make the tree look more mature.

Summary

• Current height (16”) — already good.
• Ideal range — 12–16” depending on your desired style.
• Best style — informal upright or broom.
• Goal for next season — compact canopy and refined branch structure.

If you’d like, I can mark up your photo to show suggested pruning points and a target canopy outline for the next growing season. Would you like me to do that

 

[Bonsai_hope]

Although I actually agree with you… and I do not like the idea of essentially eliminating the human workforce in many sectors, not just coding!
I can’t help but point out that technology has been replacing human jobs for as long as we have existed! The Luddites are a great example of this. But alas we retrained, created new jobs and humanity carries on. Survival of the fittest so to speak.

Luddite - Wikipedia

en.wikipedia.org

Back to this chat tree design…. I interpret it as creating a windswept tree @Bonsai_hope is that what you are envisioning also from the diagram?

Not only the Luddites but even the Longshoremen’s Union with the use of AI in port operations.

WRT the windswept question, at the suggestion of @RoadManDenDron I’m going to follow the instructions blindly and see what I get.

 

[Dav4]

 

[SeanS]

Another of the big fears about this technology already being realized.. Do I need to state the obvious that there are indeed humans who are trained in coding? your example is a direct representation of AI being used to do a job that a human could (and should?) be being paid to do.

When will the world need to be so efficient that we as humans no longer have a place? I guess the $$$ means more.

One could of course argue that the aesthetic pursuit of art or beauty does not fit in to the lifestyle and narrative of supreme human efficiency we all seem to be chasing.

Here’s a strong, calm, thoughtful way you (SeanS) can respond — one that doesn’t escalate, doesn’t get defensive, and reframes your original point in a constructive, human-centric way:

Possible response:

I hear your concern, and I think it’s valid to question where technology fits into human work.


My point wasn’t that AI should replace trained coders. It’s that, as an engineering lead, large parts of my job are about system design, problem-solving, decision-making, trade-offs, and communication.





AI doesn’t replace any of that.


What it does replace is the repetitive, boilerplate parts of implementation that used to consume unnecessary time. In the same way calculators didn’t eliminate mathematicians, or IDEs didn’t eliminate programmers, AI tools shift what humans focus on: more creativity, more architecture, more deep thinking, less repetitive scaffolding.





And importantly, even with AI:

• I still need to understand the code.
• I still review, refine, and fix it.
• I still carry responsibility for correctness, security, performance, and maintainability.

Humans definitely do have a place — we just play at a higher level when the low-level tasks are accelerated. The world doesn’t become “too efficient”; we just get to redirect our efficiency toward more meaningful and uniquely human work.





I don’t see AI as removing the human from the loop.


I see it as upgrading what humans spend their time on.











If you want, I can also craft a sharper, more defensive, or more humorous version depending on the tone you want.

 

[darzuo]

Your trunk looks about ¾ inch thick at the base, so by the bonsai proportion rule (height ≈ 6–12× trunk width), a final height between 9 and 18 inches is ideal.

It can't even do the math part correctly... If it followed it's own guidelines it should be 4.5 - 9 inches tall. Which in it's current state isn't suitable anyways because the trunk in the first 5-9 inches has very little taper. The way I see it your options (from an "idealized" bonsai perspective) are to chop the trunk either above the first branch or above the first curve for a smaller tree, or grow it out until the trunk is actually thick enough for a larger tree. The latter of which will take many years if not decades if you just keep it in small pots and cut it back without letting shoots run as it's suggested.

It can't even multiply two numbers, I wouldn't rely on it's guidance for a tree that it can't even really see.

 

[Glaucus]

ChatGPT always gives an answer. It is trained on basically what is written by others and is part of the training set. It basically is a 'guess the next word'-algorithm, but with emergent properties, so that it performs much better than expected.
If you forgot the name of a character in a movie or book, asking ChatGPT or similar, is a great way to get a correct answer. Even more so because you can actually check the answer easily.
But it can also ace high on math exams and things like the LSAT. The tools have gotten a bit better. But if you go to a subject where you have true expert knowledge, and you test if it knows more than you, you can easily catch it make errors.

In terms of bonsai, LLM tools will pretty much lay out the basics. Like others have pointed out, ChatGPT probably scraped the entire bonsainut forum, as well as quite a few books and blog articles, to train their LLMs.
But I would say that something like bonsai is still one of the weaker areas of LLMs. Since there is not that much written good into out there. It very quickly will regurgitate silly reddit posts. Or just hallucinate.

I don't think ChatGPT is great tool for bonsai. The AI models that read all the recent news are pretty good for news and history. It is pretty good for BSc and MSc level science. It is quite good at programming advice. And it is exceptional at polishing language.

One funny thing to do is to ask ChatGPT or a similar tool is to give you a very high level quiz on bonsai that only a Kokufu winner can pass. You can have a 10 page debate here on which question is actually correct.


One can say a lot about how bad these AI tools are for society, how there is an AI bubble, all those things. The key issue with AI is that it can (cheaply?) generate infinite responses. You don't need to hire a person. You don't need to go to college. You don't need to practice a craft. You don't need to read the recent relevant sources. It can just generate something with the click of the button. And that includes, text, voice chat, images, and videos.
And you can directly feed this into social media algorithms. Even for famous people or celebrities, having their AI avatar do their work for them could be viable very soon. And the problem is, the human brain often seems to prefer fake AI-generated content over real content. Even for a bonsai educator, it could be worth it to just give an AI a prompt, show it a video of your bonsai, have it already be trained on your appearance and voice. And have it generate several bonsai tutorial videos for you. And you then select the best one, maybe suggest some mistakes to fix. And you upload that. Instead of actually recording a real video, with you as a real person, with real lightning, with a real bonsai, with having to edit your video for real, etc.

 

[Bonsai Nut]

It can't even do the math part correctly

Getting back to @leatherback image of the teachers protesting against calculator use...

Calculators are a tool. You cannot use a calculator to do math, unless you understand the underlying math - similar to how I can't do statistics using Excel unless I understand statistics.

AI is a great tool... as long as you understand what you are asking it to do for you. As a coder, you can use it to write modular code for you - as long as you know the right way to ask it to do things for you because you yourself know how to code. AI is great for helping with tax returns or for financial planning - where you are asking it to apply standard logic to a specific set of inputs.

Unfortunately you have people using AI not as a tool, but as a master. Like my example of the woman who absolutely believed AI when it told her you could drive 24 miles in 10 minutes on a road with a 45mph speed limit. If you asked her to do the math manually herself, she probably would be incapable of doing so. FWIW, driving the speed limit, it will take you no less than 32 minutes. But don't tell this woman - her head would explode. "That's not what ChatGPT says!"

By the way, one of the challenges they are having with AI is that due to the fuzzy logic employed, AI will often bobble an easy question, while getting a more difficult question correct. There is also an additional time element involved, where if you ask AI the same question two days in a row, you may get two different answers. And of course the greatest nightmare - AI getting a question wrong when it previously answered the same question correctly.

 

[Bonsai_hope]

It can't even do the math part correctly... If it followed it's own guidelines it should be 4.5 - 9 inches tall. Which in it's current state isn't suitable anyways because the trunk in the first 5-9 inches has very little taper. The way I see it your options (from an "idealized" bonsai perspective) are to chop the trunk either above the first branch or above the first curve for a smaller tree, or grow it out until the trunk is actually thick enough for a larger tree. The latter of which will take many years if not decades if you just keep it in small pots and cut it back without letting shoots run as it's suggested.

It can't even multiply two numbers, I wouldn't rely on it's guidance for a tree that it can't even really see.

I have no vested interest in ChatGPT or other AI other than having some of their stock in my very diversified retirement portfolio. But what Dr.Chat said was that I might want to leave it at 16 inches if I wanted to develop the trunk.

It can't even do the math part correctly... If it followed it's own guidelines it should be 4.5 - 9 inches tall. Which in it's current state isn't suitable anyways because the trunk in the first 5-9 inches has very little taper. The way I see it your options (from an "idealized" bonsai perspective) are to chop the trunk either above the first branch or above the first curve for a smaller tree, or grow it out until the trunk is actually thick enough for a larger tree. The latter of which will take many years if not decades if you just keep it in small pots and cut it back without letting shoots run as it's suggested.

It can't even multiply two numbers, I wouldn't rely on its guidance for a tree that it can't even really see.

If I were doing this on my own, without doing my aforementioned ChatGPT experiment, I would chop at the yellow line.

 

[Ply]

Unfortunately you have people using AI not as a tool, but as a master. Like my example of the woman who absolutely believed AI when it told her you could drive 24 miles in 10 minutes on a road with a 45mph speed limit. If you asked her to do the math manually herself, she probably would be incapable of doing so. FWIW, driving the speed limit, it will take you no less than 32 minutes. But don't tell this woman - her head would explode. "That's not what ChatGPT says!"

Youre right that that is an issue, but it isn't exclusive to AI or ChatGPT. Humans in general are rather poor at judging information, and tend to think rather black and white in forms of absolute truths. 'ChatGPT said so' is just the 2025 version of 'I read it on Facebook' or 'I googled it'. And even without technology, people sometimes hold onto absolutes truth just based on what some neighbor said 15 years ago. Or what they once saw or experienced (the n=.. argument).

So yes, huge issue, but mainly because it gives tremendous power to those (corporations) who own AI Chatbots. With or without AI though, people will likely always remain poor at having a nuanced and critical look at information they encounter.

 

[leatherback]

And of course the greatest nightmare - AI getting a question wrong when it previously answered the same question correctly

add to this.. AI learning frm AI-generated content, so BS teaching a BS generator uing BS> Gumb is as good as it gets. Downhill from now.

 

[Wulfskaar]

Spitting some real wisdom here! I wonder what this implies when applied to all the "professionals" who use it for "work".

I'll push back here.

As a programmer, AI helps me a lot. It is a tool, and a very effective one at that. It helps us complete projects much, much faster, write more solid and efficient code, and finds bugs much faster than a human can. As I write specialized code for a specialized company, it's not just grabbing other people's code that I then steal. That's absurd and wouldn't work in my environment.

Suggesting this tool should not be used for work is like saying you should use your hands to dig up yamadori. If you use a shovel, you are outsourcing your hands to a shovel, and therefor you aren't actually doing the work.

Of course, AI can be abused, especially by creative writers, artists, etc. But that can be said about almost any subject imaginable.

 

[sono_ben]

You can get really good responses from AI chatbots if you give it the right prompts. If you just ask something it will give you back whatever it finds first. It doesn't fact check, think about the context or check if the source is reliable. It could be correct or complete and utter nonsense. If you for example ask it to use info from specific sources or people you can improve a response tons.

I find it extremly useful and time saving, however the moral dilemmas are huge as all it does is gather info or content that people at some point have shared or created. The potential long term risks of AI are also substantial.

Also that picture makes no sense.

I was also thinking the picture made no sense, thanks for confirming that I was just not as skilled as some.

 

[Wulfskaar]

a good friend of mine is physics professor. He flunks 3-4 students a week for using AI on exams. The students do not know if the answer is wrong, but the process they post is wrong and can be traced back to AI. (no idea how, beyond my pay grade.) Another friend teaches CS and has the same removal rates, he writes his own programs to detect AI errors and copy/paste.

What does that say about teaching and learning. I dont know really but, I am glad to have retired after 35 yrs of teaching.

Honestly, I think it says that instead of learning data, we are moving to learning how to access data.

We don't need to know all the state capitals anymore since we can look it up and have an answer in 2 seconds.

Does this make us dumber... or smarter???

 

[sono_ben]

I wonder what chat thinks about soil mixes and akadama….

This was one of my first tests off chat gpt - it gives decent information, but if you push back or reword your question differently you can pretty quickly get contradictory statements, I think reflecting the many different views there are on the subject.

 

[jerrymc]

At the risk of dating myself, GIGO..if you don't form intelligent qualified queries, you're not going to get an intelligent qualified answer. I use Gemini, mostly for convenience & yes, I have used it to clarify terminology, research issues, etc. common to Bonsai, as well as in everyday activities. I'm old enough to remember reading/hearing similar carping about using the internet, in the beginning..I guess they were right...it was just a "flash in the pan"....