AI-assisted posts / content?

[Laptop65]

This thread has inspired this: Mass Discord Ban Wave Targets Illicit OSRS Services

As we know, AI has taken off the past few years and is quickly changing the way we work, communicate and produce content (college students using AI to write their essays for them is a typical example). Should this be an allowed or potentially discouraged thing on Sythe as far as using AI to make forum posts?

I haven't made this as a suggestion as I'm curious on community + staff's view on this initially. Should 'Chat GPT-like' responses be permitted? Or should only "human" responses be allowed on the forums (Spam Forum being the exception)? I don't really like the idea of browsing + discussing a forum thread and having obvious AI-generated responses being posted, it just takes away the human element of the forums where people can post threads and discuss topics like we always have.

Obviously AI will reach the point where it can easily write / produce posts to the point you would not be able to tell it from a human response (so we could reach a point we can no longer tell if AI has written a post or not), but at least cases like the example thread above where it is obvious AI was used to write up the post could be detected.

 

It's rather easy currently to tell when a post is written by AI, but you're right in saying that we haven't taken a stance yet.

The issue is that people already use AI for disputes and reports, and if we disallowed such responses we're bound to see at least a few false positives.

I'm personally in favour of disallowing it, because frankly most of the content that comes out of AI right now is word salad at best, and likely deters more readers than it attracts.

 

[Utopia]

The obvious use of AI should be prohibited. It’s an eyesore and doesn’t leave much room for discussion.

 

Without analyzing why, I agree that this form of posting should not be allowed.
However, this is the only post I've seen from a legitimate user that was predominantly written by an AI, and not just used to correct grammar, rephrase, or give a message a different tone, or anything like that.
This person IMO has only recently discovered chatbots, or was just excited about the information provided about an actual dramatic event. I think the post was somehow decent. On a case-by-case basis, if it can add value, I think it is fine, and it has not gotten to the point where we need to make a rule about it (yet, maybe, we will see).

I would leave it as it is unless it becomes a problem.
If a staff member sees a "garbage" spam post, he may be able to treat it as spam.
But again, this is not currently an issue, IMO.

 

[President]

For the reasons mentioned above, I don't believe AI use should be prohibited.

However, its use can be factored into evaluating an outcome, as it reflects a lack of effort and commitment.

 

[paul23]

Hey, I totally agree with you. AI takes away that "human" touch that makes Sythe unique. It’s cool there's the spam forum, but overall I’d support tagging AI posts – especially in important discussions. This way we know what's "authentic."
If AI reaches the point of writing just like a human, that’s another story, but maybe there could still be a mark to indicate it's an AI post, like with tags. This keeps the line clear between humans and machines, preserving Sythe's community feel

 

I dont think it should be prohibited, but I wouldnt encourage it. The likes of chat gpt can be a very useful tool if used properly and not just abused lol

 

[Laptop65]

Based on the responses above, while it's not obviously being used widespread across the forums it would be good for the staff to take an official viewpoint on this now.

Personally I feel things like requesting a pardon should be 100% written by a human, if you're serious about righting your wrongs then you would be expected to spend time and put thought into your pardon statement and what you want to say to the staff about why you fucked up and should be allowed back. I do agree however we are going to reach the stage where AI will improve and get better to the point it will become impossible or very difficult to decipher or prove someone's used AI to write a post.

Either the staff slap it in the rules saying not to use it at all (unless in Spam Forum where nobody cares what goes on in there) or to have "limits" on where it can be used.

I think there should be a sticky in the pardon section that states "your pardon should 100% be your own written words, not AI generated."

 

[CuppaJava]

Personally I don't think AI posts should be disallowed because that's too broad and impossible to enforce reliably. I don't even think it should be disallowed in appeals or reports; If someone is not a native English speaker and uses AI responsibly and concisely to convey their message, I don't think that's an issue, for example.

The issue here is clearly the long form vapid nonsense chatgpt or other AI spits out by default, and I think that can be treated similar to any other spammy post. If someone manually, or using AI, writes a 10 paragraph appeal saying nothing, or posts vapid nonsense, that should all be treated the same. Doesn't matter if AI is involved.

If someone uses AI to condense their message or make it more readable that's fine imo.

 

Handling an entire report using AI should definately be prohibited.

 

[oldscho0l]

I feel like theres pros and cons to using A.I. to assist You . Like buddy says in his post
"Disclaimer: This post was created using historical information, publicly available reports, and AI-assisted writing tools. While AI contributed to drafting the text, the facts and analysis are based on credible sources and relevant insights, ensuring the information presented is accurate and well-grounded."

Me personally I typically dont care to use proper punctuation to the point where im being graded on an essay . That being said Im known to have idk, run on sentences sometimes..long paragraphs that can be a little convoluted . I feel like if you put in the ground work and get all the info yourself then ask chat gpt to structure it so its more readable thats not necessarily a bad thing . Poorly structured reads especially long ones can turn people off bothering to partake in your posts as its not as enticing as something more dynamic . But ya like you said if people assume its 100% gpt generated they might not wanna bother either cause why should they spend time reading what u took no time to write so its hard to say

 

I'm not against AI, because it's actually good if used properly. But the walls of text people generate with it looks like a 10 year old writing a school report which gives brain cancer to read, and they should refund all the brain cells they took from us.

 

[Ziny]

It's pure laziness not to put in some effort to be creative with your prompting skills.

 

[OhTown]

Ai is a very useful tool if used properly, I do not think we should use it to create and claim as our own or to persuade others. If we were able to enable a policy to label said ai posts correctly knowing the difference between the 2 for a fact would be a good standard.

 

[Laptop65]

Few months later in the year, AI-written posts whatever seem to be slowly increasing, example here: --------

Edit: New suggestion about users who not posting from home ip

Did the staff / community in general ever come to a consensus on this?

 

[xKyyL]

Few months later in the year, AI-written posts whatever seem to be slowly increasing, example here: Add a "dislike" button

Did the staff / community in general ever come to a consensus on this?

Who in this topic is using AI for their posts?

 

[Laptop65]

Who in this topic is using AI for their posts?

My bad it is actually this thread: New suggestion about users who not posting from home ip

 

[xKyyL]

My bad it is actually this thread: New suggestion about users who not posting from home ip

Gotcha! Yes, I agree. It is very obvious in that thread.

 

[Laptop65]

AI usage has finally been formally addressed today in the rules where it will be treated as "spam" in cases where it is being over-used or to spam large walls of "AI word salad" text.