http://dbd.game/killswitch
Can we get a ban on posts made with GenAI?
I've noticed a recent uptick in forum threads that are very clearly written by ChatGPT or a similar LLM. These posts are often weirdly worded/formatted and full of misinformation.
This should be a place where people discuss their own opinions, not asking a computer to spit out an opinion for them, and then copypasting it into a thread. Allowing these kinds of posts undermines the integrity of discussion, and is a net negative on the forums as a whole.
Comments
-
I think the posts are okay if the prompts that the user used is very well defined and not like copy paste the first thing that the ai pops out from saying "make me a balanced rework for blight".
-25 -
Well like the computer needs the relevant information to make that post. Perhaps your just seeing people who are not main English people attempting to bridge the gap with AI kinda intelligent and great use of the tool in my opinion. Also neatly formatted information is generally received better then "wall of text" so even and Amarica! Speaker like my self could use it effectively!
0 -
Also! Good news is now u can copy paste these posts into a online verifier to check a watermark for any signs of AI.
There are hidden symbols in between letters
-3 -
I agree with the sentiment, but I see no reasonable way to actually enforce this since people can always edit the chatGPT or other LLM output to remove the obvious tells.
5 -
I am wary about this simply because people will use this reasoning to report posts they don't like, and without the obvious tells (see the discussion in the AI thread about the various obvious tells) it is harder to identify right off the bat if it really is or isn't.
This... literally isn't true, and you should never use those AI checker things because they are just as prone to hallucinating as the actual AI are - because they're the same things. It's similar to how you shouldn't use Nightshade/Glaze, as its also just a AI image generator, but with all functions besides img2img turned off.
1 -
tell that to my college.
Got flagged four times this semester for AI use.
12 -
No im saying OpenAI is watermarking their AI content with their own watermark that companies,users can identify and it cant be bypassed unless you somehow make your own AI without watermark
Generally it was added to put a stop to this exact issues we have right now where users use it to be lazy, the tool was meant to be a helping guidance tool to see wider picture and not to replace the human imagination.
0 -
I think we should be pro open for AI to be used as a helping tool, but it should not be used to replace human thinking and thought process fully.
Its always good to layout your own brainstorm and then ask AI for guidance to help it sort it out and make it make sense for general public that may or may not see the vision
-2 -
Lol i majored in deep learning machine learning.
These things are dope they can help people a lot with brainstorming, calculations pretty much anything.
Its sad to see people just use it to do their homework for them or write a project
1 -
I assume you aren't actually using AI, yeah? Just based on the way I've run into your posts here, I'm honestly surprised you've gotten 4 flags!
Ahhh, okay, makes sense that its OpenAI watermarking. I was thinking of it as a whole, as I know ChatGPT and Gemini do not do anything of the sort, which is almost certainly why we see way more of those sorts of things than OpenAI posts.
1 -
They just started doing it, still in beta phase
I think its great i think people should put a credit like they always, honestly when i did my projects for uni i always included source as it was a must.
Nowadays people get sloppy and dont put any credit neither their friends or family even.
I think it will be a good change for people who dont put at the end credits "chatgpt" or other form of AI.
There is generally no shame in using AI its great its awesome, it helps spread awareness if for example someone cant properly make a message to the majority
0 -
In theory, I agree with this, but it's completely unenforceable in practice.
There's no reliable way to differentiate between generative AI and someone who just has stilted manner of writing. Or even someone who doesn't speak English as a first language and has ran their work through a translator (which is completely valid and should not be banned).
I, for one, write fanfic and have been accused of using AI before because of liberal usage of em dashes. Which is commonly cited as a "tell" that a work is AI, but the only reason AI uses em dashes a lot is because the human works it is mimicking used them first! Lots of the ways people say they can detect AI work are flawed for this reason and just end up with legitimate users getting AI accusations.
10 -
Hm. I write fiction for fun and I've never heard that. I love em dashes. When in doubt, em dash it.
2 -
I always wanted to be a writer but i just cant get my head into it, too much other life priorities and jobs i need to do.
0 -
yeah, obviously my academic writing is very different and I put, you know, real effort into it lol
Still, they've been on my ass about it because I'm an excellent writer and because I'm in engineering.
3 -
On Threads the "em dash = Ai" has come up a few times and the main responses are "people are outing themselves for not reading books" and alot of genx and Millenials saying you can tear em dash from their cold dead hands. 😂
2 -
This game has totally thrown me out of whack with it. I wrote everyday for years. Have a novel that needs revision. A couple more mapped out. About 20 short stories. I'm part of a critiquing site that I was really active in. Was also reading 50+ books a year. Now I just own a library worth of books I don't read. Took me two months to read the last one. Every time I tell myself "I'll write tonight" or "I'll read tonight" I boot up this stupid game instead.
3 -
It's a tip that circulates on social media every so often, particularly TikTok but has also cropped up on sites like Wikipedia. Allegedly, you can tell text was AI generated if there's a lot of em dashes or other "uncommon" punctutation like semicolons or ellipses. The rationale being that they're unusual punctutation for humans to include in everyday writing like texts, e-mails, job applications etc. And I think there is a degree of logic to that, however, people often incorrectly extend the advice to creative writing too, where such punctuation is actually common and always has been. Especially where amateur creative writing is concerned.
Sadly, this has lead to a a fair few comments on popular fics asking "Is this AI?", or social media call-out posts of authors being "exposed" for using AI and other such nonsense.
0 -
Right! The em dash has always been around, at least for those of us who are bookish. I was baffled the first time I heard someone say "Only AI uses dashes" lol
0 -
I'm not on social media so I'm oblivious but that's absurd. Sure I'm not gonna use a semicolon in a job app just it has its place. You can barely get most people to even place a period at the end of a sentence nowadays. Just look at the mile long rambles we occasionally get on this forum. Pretty soon all correctly-structured sentences will look like AI if we get dumb enough.
3 -
as a semi colon enjoyer and ellipses appreciator I feel called out; I always had a hunch I was a bot and my parents just didn't have the heart to tell me...
4 -
I would love to work in your library it seems like such a laid back work calm
0 -
I thought it was engagement bait the first time I saw it. I couldn't believe basic grammar/punctuation is what people are using as "tells" for Ai. 😂
1 -
There's not much to do in a room in my house haha. It's not a real library, it's my library. I own a couple thousand books.
Wish I worked at a library though. Pretty much my dream job.
0 -
I didn't realize using punctuation outed my age range lol
1 -
Oh i took it too literal, yeah library job sounds like one of those dream jobs.
Some pet item shops are also chill laid back
0 -
I doubt there's a proper way to enforce that that won't result in false flags.
I instead propose we dunk on AI takes the old-fashioned way.
Crow of Judgment
6 -
I do agree we should at least ask them to put credit: chatgpt (example)
0 -
I will be honest with you all.
Despite being described as, "The only literate engineer", I never learned how em dashes or semi-colons worked.
In a bizarre twist of irony, that is exactly how I was able to show my professors, and the department head, that my work was not AI-generated. After all, what AI would be dumb enough to use exclusively commas?
5 -
Can't believe this is even a thing people do. Wow
2 -
Ive heard this too. As someone who uses dashes (hyphens?) Ive made a conscious effort not to because of this lol
1 -
Em dashes are pretty flexible. They're great for emphasis. Almost like an exclamation mark in the middle of a sentence. It's kinda like how an ellipsis is a slow, suspenseful pause. But an em dash is a dramatic shift.
Semicolons are kinda like soft periods or hard commas. The two sentences are complete but they're so closely linked that a period is too strong. They're a little trickier to know when to use, and not terribly necessary most times.
2 -
I use ChatGPT to revise what I write, fixing sentence structures and grammar. Mostly in larger responses. I agree with you but would like to include those that use it similar to how I use it should be allowed.
4 -
Ive made a conscious effort not to because of this lol
Yes, I sometimes find myself consciously going through and removing them because I don't want people to think ChatGPT wrote it 🤣
dashes (hyphens?)
The names are often used interchangeably because of how similar they look, but technically there is a difference.
Hyphens are the shortest (-) and primarily used to join words together, indicating that multiple words are acting together in a sentence. A good example is "fancy-looking car". The hyphen tells us that both "fancy" and "looking" are describing the car. If we omit the hyphen, then the phrase could be misinterpreted as meaning a "looking car" is fancy — perhaps that's some new model of car that the writer is referring to? The hyphen makes sure the sentence is interpreted correctly, although they often gets omitted because native English speakers tend to be able to parse the meaning out anyways, even if there's some uncertainty.
En dashes are roughly the length of the letter n (–), and are used to indicate ranges. E.g. 2016–2025.
Em dashes are the longest — about the legnth of the letter m — and are used in a variety of ways. Most commonly for emphasis, or to offset additional information in a sentence, as I did in the previous sentence.
2 -
Thanks for the info! Good to know 😊
1 -
I am a generative AI hater, so yeah, ban it.
I would rather see a ranty unformatted post than a article-length ChatGPT post that can be summed up in a single sentence.
7 -
There's never been a rule or law that's been enforced with 100% accuracy. IMO "we can't guarantee we'll catch everyone" is a bad argument against a rule. Like, sometimes murderers get away with it, but murder is still illegal obviously.
Also, judging by what my teacher friends say, there are loads of students who will just copy-paste without removing the obvious signs of AI, and there the stakes of getting caught are much greater.
I have my own worries about banning GenAI. For example, I like using em-dashes more than the average person, and I like using numbered lists. Although I generally use them differently than GPT does (GPT uses them for section headers, I generally do them to enumerate a set of 1-line ideas) I'm aware my normal writing style is closer to "the GenAI style" than most. There's definitely reasons to not ban GenAI, but "it's hard to enforce" imo isn't one of them
0 -
Bots don't act up like you folks do. Probably not as much fun to ban either. Bots are the future of this forum.
-9 -
What would be the point of a forum if it’s just bots regurgitating hallucination-filled “opinions” at each other tho
9 -
No to mention the sites I've used to detect AI are crap. I've copied AI written text and it will tell me it's not AI and when I paste in something I've written from my own mind it tells me it's AI.
1 -
I imagine it'd be useful for scientists and nothing else. How long does it take for AI bots to become tribalistic.
1 -
I support this, ai needs to go the way of the dodo.
3 -
Now I really want to see two bots going for each other's throats. Each calling the other "disgusting killer-/survivor - main.
0 -
To be fair, colleges really love to freak out over plagiarism to a bizarre degree. I still remember my friend refusing to take his final during COVID because they wanted him to set up like 4 cameras and install spyware on his PC. Thankfully he still had a passing grade anyway.
1 -
This isn’t how AI watermarking works (a quick copy+paste of chatGPT output into a unicode editor will confirm it). It works by stringing words together in a way that patterns appear that are very unlikely to appear in natural human text.
A good writeup of the ways this is done can be found in the “related work” section of this research paper.
1
