Topic: Whose fault is it here? (Read 231 times)

Two other detectors that I use frequently report different results for the same body of text:

Hive AI Detector - 0.0% AI-generated
Copyleaks - 79% probability for human

This is why we have the thread rule that at least two detectors should be used when reporting a post.

Besides, as others have already pointed out, the entire reason that I found him was because of his use of the phrase "knowledge cutoff" which is only used by ChatGPT. This makes it 1000% obvious he is using AI to write his posts.

I for one am of the opinion that copy/pasting AI-generated text is akin to plagiarism -- the poster is taking credit for something they didn't write. If they clarify that they used ChatGPT to write the post and put it in quotes, that would be another story. But they don't -- they just want us to believe these words are coming from their own brain, and they clearly aren't. Seems a little dishonest to say the least.

AI detector could give a false positive, as several other members pointed out here, but the post of your [for the lack of better words] mentor here, the one being referenced as the accusation point of an AI generated text, is a textbook example of chatGPT.

It might also worth to mention that the thread being referenced here do not just take one post as their basis sample to minimize human error [i.e. "users' fault], they require each post to be verified by at least 2 of the 6 AI-detector, and there has to be at least 3 posts proven with such parameter.

So, it's AI detector's fault for your example for LMF's post, but it's somewhat less likely to be a user's fault for the post or the thread being referenced here.

I also don't understand why we even discuss the accuracy of AI detection, this 4 lines of spam is clearly copied from ChatGPT, you can argue about writing styles or punctuation or whatever how much you want but the knowledge cutoff date is the clear giveaway that not only he DID use it but also he was so lazy he didn't even check what the response was before pasting it, probably if le last phrase would have been "I lack data so I'm going to copy paste" he would have posted is the same.

I don't know what the rules are for using AI tools to generate posts and the whole deal with his alts is but one thing is clear:
- he did use ChatGPT to write that
- he was lazy to not even read before hitting post

so the whole thing of him being a pillar of knowledge or something like that is pure bs!

Right, there can be false positives, but if you'll try to go through big enough amount of posts on this forum with several most accurate AI detectors, you'll find out that nearly all posts are detected as written by human, 99,9% of all recent posts. It can also be false detected, because some little edited AI written text can be detected as human written, but well, not the hugest problem.

So when you point that AI detectors are not a 100% guarantee, we know that already. But if you have a bunch of posts in a row which are detected by several best AI detectors as AI written, it will be highly unlikely that it is just a coincidence.

And as for this exact case: how many human beings have a "knowledge cutoff in September 2021"?

Thank you for giving us this information. We have known for years that you are a brilliant inspector, and we trust what you wrote.

Of course, anyone who violates the rules of the forum deserves to be punished, and the two-month ban was a big lesson for this member, and he admitted his mistake. After this, he remained in the forum using his main account and published many valuable posts and topics to enrich the forum and benefit the members.

Frankly, I do not rule out that these AI detector tools show Theymos or Satoshi as having plagiarism content/writing from artificial intelligence. I think that most of these detector tools, in my opinion, do not provide any correct or close to accurate results.

Despite the fact that there is no rule preventing this in the Forum. Still, you can report a member by using these tools who uses AI and get his weekly payment in one of the signature campaigns that ban AI.


Source: https://sapling.ai/ai-content-detector

There is no rule that says that using an AI tool will result in a ban. Rather, it happened because a report was sent to moderators that the account violated the ban rule. I later provided a link to the fact that this account had previously been banned, and he received a two-month ban for his early farming on the forum. There is all the evidence that this account was previously part of a large farm.
https://bitcointalksearch.org/topic/m.50038879
https://bitcointalksearch.org/topic/m.55095773
However, I apologize. This account was really lucky to receive the smallest ban and unblock for only two months. As for my use of AI tools, it's just ridiculous; I don't know how to use this crap at all.

https://bitcointalksearch.org/topic/m.62908654

Sorry this is the accusation first post is made here the user who accused ulven used a known AI detector tools which been used many time in our forum I believe, and because of this what I have done is just to try edit the text users posted and see why and how these detectors works. So I think it’s all before of some commas and some words that been used alot with AIs these detectors decide if it’s a human text or bot text. Which will be unfair of course in case there is a coincidence


Well, I based my topic on this claim that user ulven got banned because of AI tool use since it’s the only accusation I ever saw against him and after few hours or a day he got an autoban.

There are few things need to break down here.

1. There's no rule or a moderator said if using AI will lead you get banned, using AI might make your post contain someone else words, but based on my observation I don't find any user reported Ulven did plagiarize, CMIIW.

2. Using one tool to detect someone using AI is still a weak proof, that's why nutildah mentioned several tools in his post. So only relying with sapling.ai to accuse lovesmayfamilis or anyone else did using AI is still a weak evidence.

Remember there's always a false positive in programming, not to mention there's will be a "fake site" claimed to be as AI detector.

I'd say case by case basis is still the best to assume someone is using AI, but using several tools might help you to make decision.

Someone reported him before, but as I said above "there's no rule using AI will lead you get banned".

Where is the post/ thread accusing him of using AI? or did they ban him without any evidence? Ok I re-read your local board, he was autobanned? Do we even have autoban for AI detected content?

Edit:  I have never seen in my life, that someone use "knowledge cutoff", who would use such a word?

Even though I am for banning anyone using AI  to farm signature campaigns, but as far as I know, forum doesn't ban such users. It's Ok to use AI as long as you mention it in the same post.

Well, straightforward a member of our local section (Arabic section) started this thread here
However, after reading the whole topic I was surprised to find out that a member of our section got banned (ulven), and he is not just any member but someone from whom I learned a lot and who shared many useful informations in our section or else where in this forum. So because of that I really had to look into this matter. What I found out about the accusation against him where he is being accused that he uses an AI tool to create  his posts which means the same as facing plagiarism accusation that goes against the forum's rules and any member should get banned because of that.

When I see the tools used to build the accusation against him I noticed something might be wrong and that these tools are also created by human beings and the same who created AI also thin thinks the same. Shortly and how I think is that, both can be detected and anyone can understand how both tools work after some uses so If there’s a detector simply because someone understood how AI were programmed not how they develop their self because there no such thing.
Anyways a simple comparison of the same text with commas and points beside another text without them will show entirely different results.

For more information that I have already gathered from the topic I shared the first time, some users accused one of our local board have shared some results accusing others of using AIs through certain tools. So we only compared some of their posts using the same tool they used to compare their own posts. Does this mean they are also breaking the rules and should be banned or is it just how the tool detectors are programmed to work:


This post is from lovesmayfamilis, is being declared as 100% fake : here  by the same AI detector used to accuse the user ulven, Im very sure that this post is 100% made by you but it’s only how the detector works…
I might be wrong in what I said but what The informations I shared are accurate and can be verified. Plus I’m not accusing anyone I’m just trying to clarify things.

So is it the fault if the AI detector or members in your opinion?