I WISH so much that it’ll be an issue in 50 years. More like 3-5. Exponentials are a b***
5 Likes
No worries!
As long as you don’t confuse me with this Oscar…
3 Likes
Who doesn’t need more Sesame Street in their lives!
2 Likes
My guess is trifoliate dormancy and ability to induce dormancy in the graft is greater then some cultivars can handle. I am assuming in some regions the graft wake up long before the rootstock do and starve or cant go as dormant as trifoliate can and starve.
Interesting thread. I liked the I don’t care answer. I have no plans to use AI but I must admit I have more interest after reading this thread. I never looked into it as it was nothing I needed. Still don’t need it.
6 Likes
A topic that I am a qualified expert (
) in! I am completing a PhD in Natural Language Processing sometime this summer (that means I started doing all this before it was cool); if anyone cares I have also actually worked in a variety of places on applying these technologies. My work has primarily been around topics in making published medical literature more accessible in bulk (structuring and summarizing documents individually or in groups), so it’s not exactly adjacent to gardening and farming but does include text generation and evaluation.
That being said, these models are good at generating good looking text. Good looking text is generally preferred by people over the median free text you find on the web. Good looking text does not have to be true, up to date, or factual. People have blind spots to things that look good but are not, particularly when they don’t know the subject already.
The best applications in my mind (keep in mind my operating bias here is in medicine) are where your correctness is irrelevant, where stakes are relatively low (this forum has a fair reach!), or where someone can thoroughly audit system performance and behavior. Is someone posting the text here (or elsewhere) likely to say “yeah, that looks about right”, or are they going to read everything it produces critically?
Presumably this forum is here for people to speak to other people. I am not interested in reading machine-generated content, particularly in areas I cannot assess (e.g. fruiting growing: I have six apple trees in pots and have inflicted some additional trees on my family).
10 Likes
I’m not in favor of using current AI for forum posts. I tried querying several items in which I am reasonably expert such as questions about pecans and tomatoes. Answers generated were moderately correct with major gaps in details. For example, it recommended Pawnee as a good pecan variety for southern Tennessee. It did not pick up the issue with Pawnee being a scab magnet. Of the varieties suggested for my area, Kanza, Major, and Greenriver are good suggestions. It missed several more that should have been considered such as Hark and Amling. It did not make a distinction between homeowner varieties and commercial varieties. Will AI eventually get good enough to provide really good answers? Yes, but will be several more years and a few orders of magnitude more complex than what is available today.
9 Likes
I think AI may be more misleading than useful on the forum. I prefer to know who is providing the data I am reading. If a member tells me he has experience, I can trust he would not take the time to post unless he really does have experience. So I vote clearly No
11 Likes
Awesome! An early congratulation than! Completing a PhD is a really great achievement!
This combination of the AI generated text “looking good” and “being preferred over media free text” combined with that fact that is does not have to be true, up to date or factual.
This in my opinion is a problem. Especially when the text gets posted without an extremely clear warning and more information like the prompt and ai engine used.
I don’t think correctness is irrelevant on this forum. (i can think of few to none places where it would be). The stakes aren’t super high here. (not as high as in medicine it seems to me)
I think the combination of the “easy to read but correctness is irrelevant” nature of current generative AI. Could make it a real problem (on this forum and elsewhere)
Where a user of this forum with limited fruit growing knowledge.
Is likely to prefer this AI text over more nuanced/correct text.
This to me seems like a bad thing.
(i suspect all of us here are to learn (more) about fruit growing. Thus have limited knowledge in at least some part)
1 Like
I’m not against technological advancements, but the whole point of a forum is peer to peer exchange of information and ideas. Using AI generated text allows people to appear to be engaging when in fact they need not even take the time to think out a response. Letting a computer do the thinking makes members less engaged in the actual conversations which means less will be absorbed by their brains.
9 Likes
Hi everyone,
One thing that worries me tremendously in respect to LLMs/AI is the potential for communities to “over-police” in respect to AI usage. What I mean by this is that the quality of generated content has reached a level where it is hard/impossible to tell with certainty if it was generated. This could lead to the situation that individuals with genuine contributions are accused of having used AI.
I come from academia (pre doc in CS but not related to ML/AI) and there one has to face the fact that many students will inevitably rely on LLMs for tgeses and papers - even if it is just to improve their writing (which works well to a certain degree). Where does legitimate usage in this respect end and how should such work be labelled?
One more anecdote regarding the usefulness of LLMs. While sometimes they can be worryingly incorrect I had great success getting help how to service an old walk behind Roteco tractor for which no service manuals are available. The results are still a mixed bag but if one extrapolates these developments its easy to imagine that in the near future the majority of responses will be reasonable and not contain fundamental errors.
All the Best,
N
1 Like
In a forum where there are so many people who take the topic seriously with great depth of knowledge and experience there is no shortage of people who can differentiate between high level dialogue and AI mish-mash of information regurgitated from the internet. The people who will be susceptible to not distinguishing between the voice of experience and the voice of AI will be the new people just getting into growing. Those would not be the people doing the policing.
6 Likes
I voted no. AI doesn’t take take account of credibility. It just takes all the rubish on the internet and distills it. It can’t distinuish fact from opinion. AI never grew a single plant.
5 Likes
In a world where alot of work is taking over by AI, im pretty sure AI doesnt grow fruit yet😛 im a fan of smart ways to implement ai.
Like the AI powered laserweeder that has 12 cameras to picture everything in its path, and sends it to ai powered weed determination computer. Which determins weeds and the laser destroys.
Or those harvest drones that use ai to determine ripeness of the fruit and harvesting them, resulting in a possibility comeback of high tree orchards.
Back to your original question about ai text, im fine with that as long as its marked as “text from ai” instead of “in my experience”
2 Likes
I had a think on this from both a user’s perspective and an administrator’s perspective (as best as I could manage). I think that a blanket ban is not a reasonable option, despite my strong preference for not seeing anything AI generated here.
A blanket ban on AI generated text isn’t going to stop someone from using an assistant from posting text; it just stops them from marking it. It won’t stop a determined user from making whatever point they want. I think, maybe one of the leaders has observations about this. The forum already has to operate on mostly good-faith participation, and the operating style switching from laissez-faire to interventionist would also mean an additional burden on leaders.
If “no AI generated text” isn’t an option, then I’d rather require at least a clear marking (again, getting this label is the important part), and ideally a comment about whether the poster could thoroughly verify it (explicitly not banning the participation). There are also plausibly good uses for it: translation, assistive technologies for writing a point, and who knows what else? The change I’d like to see is adding a point in the FAQ, requiring the label, maybe pointing to this thread, and including a statement about unaudited text (including text beyond one’s expertise) not adding to the conversation.
Those of us who don’t trust the technology at this point might just have to tolerate some inconvenience in what we read. Society does require some payment and if the biggest cost is some annoyance then I think we’re doing ok (I have good neighbors at the moment).
5 Likes
Maybe AI is for the lounge. The rest of the forum topics can stipulate (as they have in the past) only post pictures of fruit you have grown (flowers and veggies too). Photos from University hort depts. would be excellent also. I learn the most when I read about what others have done in their own back yard and from books! Many books!
5 Likes
I have a subscription to ChatGPT. To me, it seems to operate like an enhanced search engine. Like this morning I asked, "How low do temperatures have to drop top damage peach blossoms.? Instantly I got a credible answer that at 28 F for a few hours, stamens would be ruined. I did not have to sort through a billions ads.
I see no problem using the AI output, as I have on a few occasions here. I just say, "My ChatGPT says . . . . "
2 Likes
Can someone ask an AI to write a persuasive essay arguing that AI generated text should not be allowed on the forum? That sounds like a good use of AI. 
8 Likes
Me: << Why should AI-generated text be prohibited on an online gardening forum? >>
ChatGPT:
<< AI-generated text could be prohibited on an online gardening forum for several reasons:
- Accuracy and Reliability: AI-generated responses might not always be accurate or reliable, especially if the AI model is not specialized in gardening. Misinformation could lead to poor gardening practices or damage to plants.
- Authenticity and Trust: Forums thrive on genuine interactions between members. AI-generated posts could undermine the authenticity of discussions, affecting the trust users have in the advice and information shared on the platform.
- Community Engagement: One of the main goals of a forum is to foster community engagement and personal connections among its members. AI-generated text, which lacks personal experience or emotions, might not contribute effectively to community building.
- Spam and Manipulation: If AI-generated text is not controlled, it could be used to spam the forum or manipulate discussions, potentially drowning out genuine user contributions and derailing topics.
- Legal and Ethical Concerns: Depending on the content, AI-generated text might inadvertently violate copyright laws or spread biased information, posing legal and ethical risks to the forum operators.
Prohibiting AI-generated text can help maintain the quality and integrity of the discussions on a gardening forum, ensuring that the information remains useful, accurate, and engaging for human participants. >>
10 Likes
Nicely done, Chat GPT… I’m convinced! Some of those weren’t even raised here yet.
6 Likes