Topic: AI Spam Report Reference Thread (Read 21733 times)

Then we need to agree to disagree, I wouldn't say that simply because someone use AI to search through some topics or article rather than just surfing through Google search means that he/she doesn't value this forum. For me AI in this case ChatGPT is just a much faster search engine, for example one of the hottest topic recently, 'Trump winning US Election' I can do google search or ask AI the same thing 'What is Trump policies related to Bitcoin and how it will affect Bitcoin?', if I do google search, I will need to pick among dozens of articles and news with their clickbait headlines, read the whole article, make a conlusion about the main point of the articles, go to the next 5-6 articles, repeat, until I found enough resource, then rewrite and rephrase those conclusions and main points, add my opinion or make a counter argument, then after at least 45-60 mins I have a post.

But with ChatGPT, I can ask the same thing, GPT will make me a long paragraphs of curated articles from google search, and answers my questions, then I can pick some points that I think is relevant, check on google about those specific points to verify, rephrase and rewrite them, add my opinion or counter argument, I can make the post in just 20-30 mins.

I mean value this forum and my account, but I also value my time.

All that being said, if what I have done as I explained above is violating any forum rules, then maybe I deserve to be tagged.

I wouldn't say your statement is true, because if it were, it would never occur to you to use AI for anything in your posts. We can argue about this for days, but all these AI detectors simply can't be wrong, especially those that are at a higher level than these free ones that are mostly used.

The solution is simple, forget about AI tools in the context of forum posts.

I think your heart is in the right place. Just don't copy/paste stuff from ChatGPT, at all. That's the best way to avoid getting mentioned here. If you do have to post it, make sure to credit ChatGPT, even if only as a reference.

So, ok, as I understand from your post here here,

First, an AI generated text that is messed up so much buy the user will only be detected with supposedly more advance and better AI detector such as Originality and Winston,

Second @FinneysTrueVision accuse me that I generate the post with AI then only change the punctuation on my text.

So according to the first statement, shouldn't my text be easily detected as AI generated by even a simple and common AI detector? and also the result on the more common AI and the Originality shouldn't be too different because my text is not much altered.  

I haven't gone through the reported posts, but @FinneysTrueVision claimed that your text appeared 100% AI with minor corrections in punctuation. Thus, it's fair to assume that some of your posts are completely AI-generated and that you're not simply looking for ideas to write but to copy and paste the easy way. We're not cherry picking the AI detectors; I came up with Originality and Winston in this thread, which were the only ones that detected two members of the Stake campaign, and the first one was completely off the radar with the simple AI detectors we used because of the sophisticated way he messed up the text, so much so that the detection software couldn't identify that many grammatical mistakes as AI.

I value this forum and my account,



I don't know about robotic, but I am not native English so basic error should be understandable, and for the advance vocabulary, I guess its because google translate that I use on some words or phrases that I don't know how to put it in English, but I would say that from my posts that you use as an example, I barely see any sophisticated word. And I do agree that using just one source is unreliable, so I really don't know how to deny this, also doesn't it mean that those AI detector is unreliable because they contradict each other, and I mean the difference is huge, one with 3% AI, the other with 30% and then the other one 97%, for me that will be very unreliable to accuse any user using that data. Other than that, if you said that one is more reliable then the other just because it shows higher percentage of AI, seems like cherry picking.

It is unreliable to only trust a single detector, here are the results from a few other detectors.

Quillbot.com - 30% of text is likely AI generated
Copyleaks.com - AI content detected
HiveModeration - 96% likely to contain AI generated text


There might be some truth in this, but your posts seem unnatural and robotic that I am led to believe that even if your posts aren’t entirely AI-generated you still rely too heavily on AI assistance. I also find it hard to explain how you can write posts that are seemingly professionally crafted, using somewhat advanced vocabulary, yet there are very basic errors like adding an unnecessary hyphen or leaving out a preposition. That is very typical of users trying to hide the fact they are using AI.

Well, at least you replied here, compared to most of the users who are being reported here, you're taking accountability. I don't know what exactly you're doing and I'm not going to go through your posts. ZeroGPT isn't the most accurate, Originality and Winston are a lot more reliable because they're also for academic purposes. I personally would advise you to refrain from using AI again, I understand if you used it to gain some insight or ideas, but it's generally frowned upon, we're looking to have a discussion to see other people's opinions and ideas, not AI's.

You're the only one that came forward, perhaps you should be given a second chance, provided that you change your writing habits, I'll also wait for other members to express their opinion.

I don't know how to disprove this other than just providing screenshot from ZeroGPT, I am not subscribing to originality,



I do often ask AI for some topics but I don't just copy and paste, I rewrite and add some other info I found from google also verify the info gave by ChatGPT because as far as I know AI data is not realtime and sometimes outdated, I don't subscribe to ChatGPT or any other AI, and from what I know, AI won't write such phrase like 'In my Opinion' or 'As far as I know', but again, I don't know how to deny accusation, so feel free to decide.

This is another user who is making very minor changes to AI generated content to make it harder to detect.

icalical


Sapling.ai - Their modified post has a result of 33.7% Fake. When correcting a misplaced period and a comma, the result is 100% fake.
winston.ai - 0% human



winston.ai - 0% human
originality.ai - 57% likely AI on their modified post, 97% likely AI when removing a hyphen they added in.



This post also had some modifications such as removing and replacing words in the second paragraph, but even with these changes the detectors still gave a very high likelihood of being AI generated.

Sapling.ai - Fake: 99.3 %
originality.ai - 98% likely AI-generated

It's not really that big of a hassle to have an AI app to write for you, you simply quote the post you want it to reply to and it automatically does it for you. A few additions here and there and you can adjust your reply to your preferences, then pass it to a humanizer and you're good to go. Do that for 5-6 posts throughout the day and you've reached your quota. On the other hand, you might be looking for quite a while to find a worthy post to reply to, I'm personally a little more picky, but AI almost automates this process.

It is, but they still do it, and I'll tell you the probable reasons for that.


There are two very probable reasons why people use AI models to generate posts using AI and then use tools to humanize them.

1. They don't have the brain to think and come up with constructive points themselves, and they believe AI can do much better than them.

2. They are not very good at writing English, and they believe their posts would have more weight if they were in better English. They use AI to generate perfect posts.


They are not; even if they manage to get a slot somewhere, they are dropped when caught and reported doing this. I have reported campaign participants myself, who were then removed.

Is using Ai and then finding a way to disguise the posts from looking like it is from an Ai not enough hardwork for those who do that? Why can they not keep to their own writing style instead of passing through all this Ai thing. Anyone who is lazy or lacks confidence to make their own posts and contributions from their brain should not be enrolled in a campaign. They do no deserve it.

Good question, That user was in the campaign long before I joined management. Lauda was out of DT for too long, and most of these tagged users became free again and were able to join campaigns. Lauda's tags became visible again a few months ago, and I can't explain how he went under the radar with the second red. The new rating team is exposing them, which is a great; but, these folks are using AI humaniser tools to make it tough for AI detector tools. My team and I had been watching that user for several weeks and wanted to make sure we got it correctly.

I asked Ultegra to help scan that user profile because he is quite good at it, and the results verified that I and the team were right all along. Appropriate decision will be taken as we always do.

I will just ask the question why did it take so long? The user in question has negative tags since 2017 as a spammer, sig campaign abuser and account farmer - I wonder what he even did in any sig campaign? I don't doubt that there are more similar ones, it's just a question of how deep someone wants to dig to find them.

Thanks to @Igebotz, another Stake user has been identified using AI, a large number of his posts also appear AI generated on Copyleaks as well. Here are a few examples. Screenshot evidence can be found on the link below.

Screenshots

User: Legentbtc

Post 1.
Copyleaks: AI Content Detected
Winston.ai: 100% AI
Originality: 81% AI

Post 2.
Originality.ai: 100% AI
Winston: 91% AI (9% human)

Post 3.
Winston.ai: 96% AI (4% Human)
Originality: 100% AI

Post 4.
Winston: 93% AI (7% human)
Originality: 100% AI

Come on give him a break [up to you though].

I had a chat with the user, he is new to the environment. English is not his native language too. So he said he used AI. I suggested him to edit texts to make them meaningful, correct sentence formatting then to post. It's not always easy for local speakers/writers to create English text.

The user is representing a business so it's not that their goal is to spam the forum but to communicate with potential customers.

Consider it a vouch from me for them.

I noticed a suspicious user today, who has also recently changed email address, perhaps someone else could also take a look, I don't have any more credits left on Originality and Winston, so this is the only one I could check. Only few of them look suspicious and not all of them. The user has literally thousands of posts, more than most established members have.

User: kotajikikox

Post 1.

Winston and Originality results:

In your image, you just took 1 sentence from the post and pasted it 3 times. The sentence itself is obviously NOT AI because AI wouldn't mistake the word "proof" with "prove", which is what its supposed to be.

As a reminder, please adhere to the following guidelines, which are outlined in the OP, before making a report:

- post 3 instances of posts by the same user, where
- at least 2 detectors give a result of 70%+ AI

If a user has previously been identified as using AI in this thread, then its OK to include only 1-2 of their posts in a follow-up report.

Kindly go through all my posts and you will understand that my use of English language is quite polished, I don't have a need to use AI to enhance my grammatical structures. I read a lot and knows how to structure the contents of my posts, even though I can't claim to be perfect. So I can assure you that I'm an original person and every posts that I make on this forum, including this one are constructed by me without the aid of a second or third party.

Thanks for pointing out my mistake of misspelling "Gensler", remember that nobody is above mistake. I will go to edit and correctt it immediately.