Website & Forums Discourage the usage of AI to produce suggestions

AI is definitely used on the forums by people of varying degrees. Sometimes it's used to rewrite ideas if someone's not fluent in English or has a writing disability, for other times the OP just uses what an AI wrote up when asked "make a suggestion to improve wynncraft" and passes it off as their own. It's actually not that hard to tell apart AI posts from genuine posts. You'll find the AI uses less diction variety, it has a much more rigid sentence structure, and also follows a formula for writing posts. The biggest giveaway it's AI though is when it hallucinates stuff that's not in the game even when it already is or already is in the game but it thinks it isn't (as an example- I asked ChatGPT to "make a suggestion to improve wynncraft", and it thought world events and festivals don't already exist).

 

...you know OP is on the team that develops this game, right?

 

I have played this game. The dev team, bless them, are not what I would describe as omniscient.

 

There's a reason it says 'discourage' and not 'scour and expunge McCarthy-style'. This thread doesn't come without precedent (though funnily enough, that thread is gone now; "Allow All Classes to Use All Types of Weapons"), and the original poster of that thread even admitted to its use when the AI used created a unanimously meandering suggestion that hallucinated aspects and restrictions of the game that do not exist. Sure, you can't catch them all in a more general sense, but when having to create a substantial argument for a game WITHIN a game that is as specific, dynamic and in-depth as this, LLM-churnings will be as obvious as the occasions they had already been detected in.

 

Well, I got told off by OP once for "necro-posting" by participating in a post that was on the first page of results for general suggestions. There isn't enough activity on this forum for AI that's nonsensical to be weighing things down, and if the AI post can be understood than other people have made good arguments here about it being an accessibility tool. My argument is still that this it's a bad idea to witch hunt AI writing because it would be impossible to develop guidelines around detecting AI. The people who think a post is AI are of differing minds as to what criteria means something is AI, and I haven't been convinced by any of them.

 

And I already made my argument as to why it doesn't actually make anything more accessible when it creates more work for the end user to do to begin with. Curating, editing, fact-checking, which will be even more tedious if you have a disability you used AI to compensate for. Take your grudge with Deusphage to another thread, I'm the heel of this forum now.

 

have you considered that the reason you cant find AI posts is because the 4 AI posts made this weekend have all been taken down?

 

My only grudge is with the idea proposed. Pretty obvious considering that I'm criticizing the idea.

 

Then just mention that you think posts here are infrequent without the little jibe.

As for your previous point, you're right; there are no good criteria and cases will vary, because the AI is inconsistent and doesn't really have the extractible experience actual players of WynnCraft have, let alone anything that isn't written down, and it will never really know what two concepts to associate together outside of guessing based on a percentage of confidence. It will never gain that experience outside of piecing partial pictures together through scrapings of numbers and text. It can't play WynnCraft nor form its own conclusions of what it's like to play it. So it's not going to be very good at making meaningful suggestions for it without significant input from the end user that would have been better spent gaining experience or researching to support their point, rather than justify the word of something that spoke for them.

 

I think you overestimate the competency of AI to write suggestions for wynncraft. It has very little clue what would actually make for good features or changes, and always lays things out in the same format LLMs use.
I know for certain that some of these posts were indeed AI because when it was brought up, the creators would sometimes admit themselves that they had used generative AI to produce the content of their posts.

 

I don't have anything against using AI to improve writing that already exists (well maybe a little), but my problem is that it almost invariably puts down a bunch of nonsense. Sometimes it's just details and specific examples that are nonsensical, sometimes it's enough that I have to wonder if the person who is ostensibly behind the suggestion has even played the game. This is even worse in the questions section, where I have seen outright nonsense responses to questions. The suggestions is just annoying and at worst just ruins a chance at a really good post, but questions is really annoying and actually potentially harmful (inasmuch as misinformation about a Minecraft server can be).

I hate the irresponsibility AI chatbots foster in spreading nonsense and misinformation as opposed to actual, factual information and real views and opinions. Yes, it's just a Minecraft server, and if you make a suggestion that's totally nonsensical about how some feature that's in the game should be added or something like that, it doesn't really matter. But at the same time, don't we owe it to the people around us the minimum decency to use our own words, thoughts, and experience when trying to convince them, instead of having some chatbot hallucinate up a reason in order to get... likes on a Minecraft server forum? Upvotes on a meaningless poll? More cynically speaking, a bit of testing and feedback to improve their chatbot? It rubs me the wrong way. Maybe I'm a crazy internet boomer on this, but it just feels disrespectful.

I don't wanna sound super rude to people who are just trying to use it to rephrase something they already wrote, or who have trouble putting something into words, but I'd rather read a massive jumble of terrible wording of a good idea than a masterfully composed paragraph of total nonsense. If someone's actually doing the work and just has it tidy up their work, I think that's different from just telling to to make a suggestion about <topic>. I don't really mind the former (though I'd rather hear it in their own words), but I wouldn't complain about mods removing the latter.

 

It's harmless enough, but it does come off as a bit overly-precious of a draft in the best-case, and in the worst case distrustful of others' ability to meaningfully engage with them, as if they would be misunderstood completely or that no attempt to communicate would be made if there was a misunderstanding. (Not to say that's an unwarranted concern these days.....)

It's important to remember that these are just posts on a forum. Unless you're in 'Your Work' and you're typing a novella or two, with possibly the worst BBCode experience available on XenForo (WHY CAN'T I ITALICIZE ANYTHING? I TRIED IT JUST THERE), there is more incentive to speak as yourself than there is to type 'perfectly'.

 

For example the coop lootrun thread was AI, we all roasted it, OP reuploaded it as a word salad, and we all treated it well.

 

Small inaccuracies like that are things that can be buffed out if it ever does end up in production. I think the type of character you mean to depict in your thread comes across just fine otherwise. Anything in this forum is in its draft state, where the most important thing to portray is the outline, the concept, the spirit, of the final product.

 

See:
1: Yes, you can 'just learn [medieval English]'. It's not *that* hard. What you're considering 'medieval English' is probably not even that old; the KJV's in the public domain, and you can find similar stuff elsewhere. You could have looked up resources on how to write similarly to medieval English that weren't AI.
2: AI's not going to produce anything good anyway. ChatGPT is a terrible writer. Remember that there's nothing generative about generative AI; it's a data aggregator. It knows nothing. Whenever you ask ChatGPT to do anything, fundamentally, all it's doing is guessing what the answer will be.
3: If you're suggesting something down to the individual lines of dialogue, either you're going into too much detail or you're a CT. There's nothing workable in suggesting things to that level of detail, as Deus has said; you don't *need* to know medieval English because you shouldn't be writing it out in the first place. It's not strictly relevant - that's a flaw unrelated to using AI - but it's important to remember.

My personal thoughts on acceptable use-cases for LLMs:
- Automating boilerplate tasks with zero creativity involved (some programming elements)
- Analysis (mainly in science and medicine)
- Countering other LLMs (if a prospective employer is using AI to vet your resume, write your resume with AI)
If you think that your writing skills aren't good enough, you need to improve them. AI is a terrible writer, and using it won't improve your skills at all. One way to improve writing is by practicing. This is something you do whenever you write literally anything and aren't using an AI to do it. If you write more, you'll be better.

 

i just wrote the things and said to the ai to add the mediavel english grammar

 

This should apply to all posts on the forums, not just for those in here. The whole point of the forum is to connect, share, and interact with real people with their own thoughts and ideas, and AI defeats that purpose. As others have pointed out, the excuse of, “AI helps me write better” is garbage and if you’re using AI to the extent that people can tell it was made by AI, you’re not even getting good writing from it anyway.

If you can’t even be asked to write a post, why should anyone bother reading it then?

 

it is rare that any suggestion in need of "badass dialogue" is reasonably ever going to be added. we're not looking for quest ideas.

 

I haven’t been on the forums for awhile,, but this is a thing now? Using AI for YOUR idea is nuts imo