0 voters
A few times if seen people use AI to generate reply’s/text to topics on this forum about fruit growing.
I personally think these should not have a place on the forum.
AI (at least in it’s current form) is a word association algorithm. It reads loads of text. And based on these texts starts writing sentences 1 word at a time. Guessing each next word based on what word often follows the last word. (in the scanned texts)
This means it basically tries to copy/combine a lot of scanned text from similar subjects. Without understanding any of the text or any reasoning.
I think we should not have this on the forum. Or it should at least be clearly marked as AI generated. Giving the original prompt/command that was used to generate the text.
I started this poll. Because i was curious what others think. And also as a place to further discus without going off topic in the topic where this came up.
post 25 and post 32 and some onwards from there.
I voted other. I think it can be used in the lounge if people find a utility for it. I think this being a human experience forum it’s best left for very few instances where maybe it’s humorous to see what AI has to say on a topic. But I’m more along the lines of it has not a lot of business in most posts.
How on Earth do you propose to police that? I’m not a big fan of AI generated content, but I don’t see how we can realistically ban it.
a very valid point.
We could put a few sentences in the forum rules on how we would like people to not use AI. Or how they should mention it if they do.
In the end it will be hard to police. However so is trying to police people lying on this forum.
If i wanted to, i could troll and write lies as reply’s to peoples questions. After a little while people would catch on and hopefully some actions would be taken.
I imagine we could do the same for AI content and the people that post it. If the AI text gets so good that we can’t differentiate them from experienced users on the forum. Than the problem has solved itself. Until than. I would like to see no/less AI generated text on the forum instead of more.
But that’s just my opinion
AI generated text for things like plant care is just too unreliable even when it sounds very plausible. I would prefer for it to be against the rules (like talking politics), but at the very least a requirement that it be clearly flagged as AI generated with a warning to not necessarily rely upon it.
In the lounge on topics unrelated to plant growing, it seems more like something that could have a place. E.g., generating funny memes via AI seems ok to me.
I think the fear of AI is based on a sci-fi version of AI that people have in their mind, but that’s not what is currently out there. Generative AI is just the ability to analyze a large of amount of data very quickly and that can be quite useful for research purposes. But like any research it’s only as good as the data that’s being looked at. Bad data in, equals bad data out. No different than a human doing their own research and taking something from Wikipedia as fact without checking deeper. You may have seen the article recently about the City of New York and their citizen chatbot giving out wrong answers because they didn’t clean up old policies, so the AI was just reading and spitting out outdated answers.
AI is like any research. If you intentionally use it you should cite your sources and take the time to fact check the results. It shouldn’t be considered an authoritative data source.
I say intentionally though because, it’s very likely in the future that people will re-post AI generated content without realizing it’s AI generated content. That’s true across the entire internet. At some point you may cite a source and not realize that content was generated by a machine. If Universities and Governments are all using it, and they are, then that’s inevitable at some point.
There are other sites if people want to chat with an AI and get entertaining consensus speech.
I agree in large part with your post. Some exccelent points. And this topic isn’t meant as a “fear: AI it’s going to destroy the world” thing.
It’s more to discus what we want this forum to be. related to AI.
I disagree here. AI (in it’s current form) does not really analyze. It does not understand the meaning of words. It just guessing next word. based on associations it scanned in read texts. It doesn’t actually understand the meaning of those words. Just how often and close together they are used in the texts it scanned.
Here is another difference between AI and “research”. The AI doesn’t only give “incorrect output” because of bad data. But also because it doesn’t understand the data it used to guess a response. There is no understanding of the question or answers real life meaning.
I think there is a difference. Because the human being understand the real life meaning of the words. Even if the word are from an unreliable source. The AI doesn’t. The human being also often posts a source. The AI doesn’t. (if it did, it would be a lot more useful. Since than you could verify the answer with the source(es)
I disagree here.
Research is “creative and systematic work undertaken to increase the stock of knowledge”.[1] It involves the collection, organization, and analysis of evidence to increase understanding of a topic, characterized by a particular attentiveness to controlling sources of bias and error.
The AI fundamentally does not have an understanding of the text it uses to formulate answers. There is no increase in understanding.
I’m not saying AI is bad. Or useless. But it is fundamentally different from a human responding at this moment. Especially on a topic of which relatively little has been written. (like rootstocks or fruit growing)
It’s currently really good at mimicking human writing. But it’s quite (extremely) bad at actually understanding the meaning of words. And thus it’s accuracy can be very bad on more niche topics. Where it can’t easily “copy” an answer from a scanned text.
You can make the EXACT same argument for humans. We don’t have an adequate understanding of language to say DEFINITIVELY that we aren’t doing the same thing. You’re just rearranging words you’ve heard before, not ACTUAL understanding (no true Scotsman fallacy)
I’m not saying I agree with this view, but it’s very clear that these frontier LLMs have actual models of the world that they construct and reference, not some word soup
Thank you for raising this issue, @oscar. I voted “other”, because I’m kind of undecided between “no AI” and “only if clearly disclosed.” Like @jcguarneri said, I don’t know how you would enforce a ban, but I think it’s good to have a community standard that we value personal experience, practical knowledge, and responsible sourcing of information, and current generative AI doesn’t give you any of those things.
I’d also venture to say that most of us come here because we enjoy connecting with other people who share our interests and enthusiasms. Interacting with a bot is not going to give you that.
an interesting point. And i love the discussion this gives.
In a theoretical “experiment” an AI might come close to an paralyzed human who’s never “gone outside” and “saw or touched a plant”. But extensively read about them. They both give an answer on read text and their interpretation of the text.
However. Users of this forum actually saw or touched a real life fruit tree. Know the “real life meaning” of the word rootstock. Not just the context of where the word is often used in written text. There i think they (still) differ from the AI we currently have.
you could argue that our current AI has understanding of “text” but not understanding of the “real life” meaning or the things referred to in that text.
These AI’s do have “understanding” of sentence structure and language. But i meant that it does not have an understanding of the “real life object” linked to certain words.
Some great points! It’s such a fascinating subject to discuss. The moderators might disagree at some point though. 
Blockquote I disagree here. AI (in it’s current form) does not really analyze. It does not understand the meaning of words. It just guessing next word. based on associations it scanned in read texts. It doesn’t actually understand the meaning of those words. Just how often and close together they are used in the texts it scanned.
I disagree with this part. AI by it’s very definition does analyze data and is designed to make predictions based on the data. Understanding is whole different issue, but honestly probably more of a philosophical one. What does it really mean to understand something.
Blockquote I think there is a difference. Because the human being understand the real life meaning of the words. Even if the word are from an unreliable source. The AI doesn’t. The human being also often posts a source. The AI doesn’t. (if it did, it would be a lot more useful. Since than you could verify the answer with the source(es)
I think this isn’t true in many cases either. Citing sources seems to be a rarity these days. Unfortunately.
Blockquote The AI fundamentally does not have an understanding of the text it uses to formulate answers. There is no increase in understanding.
Again I agree that it’s not understanding, but it is analyzing and presenting you with more data than you could have analyzed on your own. So it’s increasing the understanding of the human using it, I think that’s one of the points.
I generally am opposed to using technology in my own garden and orchard and it wouldn’t make me sad to see AI go away all together. But I think on a forum that’s possible because of the latest technology, it will be very difficult to keep out AI which is one of the latest technologies.
I’m a realist in that sense.
Also I think that this forum is built on discourse (correct me if I’m wrong on that) and discourse uses many AI tools to improve it’s forums and make them function better. There would be some irony in disallowing AI for forum users when the forum technology itself uses AI.
There are times when I feel like boycotting products that use AI, and then I realize that everything uses AI these days if you dig deep enough. How idealistic do I want to be?
I much prefer original thought even if it is wrong. That is part of the beauty of being here. This is not my strong point at all. And I appreciate the opportunity to learn.
excellent point.
Maybe “allow” was a poor choice of words. Maybe it should be.
“should we encourage users to not post AI generated text”
Depending on how broad a definition you want to apply. You could call a PID controller (in a seedling heating mat or your home thermostat) a very basic version of AI. Here the algorithm is still understandable to humans.
But that’s a very different thing from AI generated text. That’s why i purposefully choose the title
“Should we allow AI generated text on this forum?”
and not “Should we allow AI on the forum?”
I’m not fundamentally opposed to AI. AI has the potential to do a lot of really amazing and useful things. But currently text generative AI is really good at mimicking human writing “style” but not great at giving actually “correct” answers. Especially on more niche subjects like fruit tree rootstocks for example.
This disconnect between a “seemingly well written and confident response” and the high chance of errors in the text. In my opinion make it a “bad” thing on this forum. More experienced members probably will spot the errors or unreliability in some AI generated text. But newer members might not.
You’re right. I think that’s where a good policy will go a long way in setting expectations. Using AI to generate your post is a very different thing than using AI to help gather research. I personally don’t have the time for either, but I wouldn’t fault someone for using AI long as they’re not passing off AI content as their own. In that way it’s no different than using Google in my opinion. The technologies to post words that aren’t your own already exist without AI.
If this is really how it’s done, then it is just an average answer and an outdated answer. Not something with new or real insight. In any field it’s the few experts that are the real authorities. And they advance understanding. Average on the internet sucks.
