Artificial Intelligence on the Forum

[logfiles]

I have to say this one is true. Detecting AI generated posts could be easier to say but when you are in the process, it might be harder than it is, most especially if you simply check his current posts without going deep into his past history. Probably the reason why these AI users are even motivated to continue what they’re doing because it will be harder for them to get caught. And maybe if they get caught, that would mean an extra work for this forum admin. Something that I can really say detecting AI post is really hard and somewhat tricky.

But there have been examples where one could clearly see that there has been intentional usage of AI to generate texts. Should we just keep ignoring such cases? How different is that from plagiarism that has been demonized in the forum from almost as long as it has existed?

I think the forum administration needs to come out on this, otherwise if we condone it, then plagiarism should also be condoned since it's also "hard to detect" at times

[Filippo Spina]

Spammers and scammers are going to be top AI users, if they are not already.

It's because spammers and scammers don't realize that AI is supposed to just be a tool to augment something, not entirely replace it.

Like, Chat GPT is really good at telling you how to do something simply, or making a guide on how to start a fire. It's not going to, nor is it supposed to, start the fire for you IRL.

Idk if that's a shitty example, but yeah.

[Vod]

AI is supposed to just be a tool to augment something, not entirely replace it.

I disagree - a lot of development in AI is to replace the human, such as in warehouse or other repetitive/dangerous/expensive work.

As a young mental health professional, you would view AI as a tool to help you, because your job will be one of the last to be replaced.   

[LoyceV]

Spammers and scammers are going to be top AI users, if they are not already.

I'd expect regulators to act on this, but it's going to take many years before they do something. And given the great "success" of having cookie warnings everywhere, I don't have high hopes. Making the AI-manufacturers liable for the problems caused by their software would probably end this, but that's "killing innovation". Add a strong lobby, and I don't expect improvements.

[Quickseller]

no person in their right mind would just go and ask chatgpt questions and copy and paste them into forum threads just for free.

It's the modern version of shitposting account farmers. Some even manage to earn Merit with it.

There is a lot of money to be made via LLMs (which is really what the OP is referring to), and that is not from sig deals. The creators of LLMs want their models to be relevant and accurate, so I somewhat understand why some might use the forum to create LLM-created posts, although there are probably some ethical issues in doing so.

Anything a LLM says is not plagiarism. If there is a topic that I find interesting regarding something I don't know a lot about, I might read articles from what I believe to be reputable sources, and I might paraphrase information that I find across several sources in my post. That is essentially what a LLM does.

[LoyceV]

I might read articles from what I believe to be reputable sources, and I might paraphrase information that I find across several sources in my post.

Any university would require you to credit the sources, and if not, it's plagiarism.

[DaNNy001]

Is grammarly and other tools to improve post quality is part of this AI tools? I’m using this AI tools sometimes when I’m on a device with this extension which I use to correct my grammar and improve my post. Sometimes it change my sentence construction to a more decent format.

I’m confused whether this tool is considered as AI or not but I'm sure that it’s different approach compared to chatgpt AI because my thoughts is the basis of the improved post.

I do not know how to answer this question but I would say after you have used grammarly to help your sentence construction, run your text through a AI checker software available on the internet, if they turn out to signal that it is AI written, I would advise that you should simply indicate at the end of the text that grammarly was used in helping your sentence construction. Because if any other user runs your texts via the AI checker and it turns out to be 100% AI, how do you explain to them that it was grammarly that you use for sentence construction and not chat GPT?

Anyone who often uses several tools to check AI texts can already visually predict whether a post was created using GPT chat or not. I'm not talking about guaranteed detection, but such posts most often attract the eye and beg for verification. If you're talking about grammar and spell-checking tools, then your text is unlikely to be detectable as AI-written text. I see a lot of examples, and those posts that people write on their own will never (or in rare cases) be identified as AI. Therefore, there is no need for clarification.

There is the topic, and there are quite a lot of examples of how people catch spammers in GPT chat. Moderators also use their own verification methods; otherwise, how can we think that some reports are good and some remain unprocessed?

But those cases when we mark spammers as “accounts using AI” have virtually no effect on the spammers. Having received the tag, they continue to write in AI texts.

It really is like fighting windmills.

Thank you for this clarification. I usually have a doubt when using grammarly since they now offer text improvement which rephrased a complete statement based on the format your desire by changing the tone and expression of your original statement. This makes me confused because this might already a borderline AI like tools which I might unknowingly using.

But one thing is for sure that I construct first my first and just improve using the tools to make it more appealing to read by correcting the grammar. This comment give me confidence. Thanks again

I also was using the grammarly as a supportive measure to atleast straighten and make my grammar a little bit readable but the whole chaos with AI generated or correct post made me let go because it got to a particular period when I actually felt that even the use of grammarly could be classified as AI aided post or write but I know all the writes up are made originally by me, it's just that the grammarly too helps me rearrange them to put the right punctuation and grammar set up in the right order.

[Cryptomultiplier]

Ordinarily, it's an act of cheating on others by using AI to generate a topic and laying claim of ownership to it. It could make those other members that are constantly driving in personal  effort into making quality posts get to feel as though their efforts is not making progress or improvement whenever they have to compare an AI generator poster posts to that of theirs which was done organically of their own effort. That's just one of many reasons I don't support the use of AI in the forum for discussions.

But if the administrators would want to absorb the use of AI in anyway, then I'll suggest a dedicated board is created for all kinds of AI generated topics no matter the category or section the topics should just be posted into that very board in as much as it's AI generated. And it won't need the poster to indicate if it's AI generated or not, just by having it there common sense will tell it's AI generated.

I buy the idea of having a separate board created for AI generated essays, even though that it would create a divide because any topics already created by human intelligence may have a different ring to it if created using AI.
I read here where someone suggested that users who make comments or create post using Ai, should indicate at the end or leave something like a footnote to state it was created by AI.

In all, the essence of having a community like this where we as humans from different regions in the world can share real experiences and ideas that isn't AI generated, is simply beautiful and I have always loved to partake in the reasoning of others as much I share mine, mostly those of whom I have neither seen or met or belong with in same age bracket.

[Pmalek]

Anything a LLM says is not plagiarism.

Let me correct that for you. Everything an LLM or text generator writes can be plagiarism, depending on how you use it.

If there is a topic that I find interesting regarding something I don't know a lot about, I might read articles from what I believe to be reputable sources, and I might paraphrase information that I find across several sources in my post. That is essentially what a LLM does.

Acquiring knowledge by reading what others have said on a particular topic and then explaining or writing about it the way you remembered it is not plagiarism. If it was, most of the stuff we say about Bitcoin could be classified as plagiarism. A simple question, like explaining what Bitcoin is can't be answered without saying something that other people haven't already said in the past (most probably).

Plagiarism is often about intention. If you find a research on hardware wallets and make it sound like you made the research, it's plagiarism. You can't write about that on the forum by saying that you conducted the research, you talked with experts, you spent hours configuring and working with the devices, if it was someone else who did it.

I also was using the grammarly as a supportive measure to atleast straighten and make my grammar a little bit readable but the whole chaos with AI generated or correct post made me let go because it got to a particular period when I actually felt that even the use of grammarly could be classified as AI.

Grammarly doesn't generate text on its own. It corrects the grammar of existing text you feed into it. You should be fine using Grammarly. No one should get into trouble for doing extra work to make their posts better, especially if English isn't their first language. It would be unfair if the admins punished users for using Grammarly.