AI generated content by newbies

[mirakal]

I don't want to automate anything related with AI, and don't fully trust any tool that is checking for AI generated posts, but using multiple tools is making those chances much higher.
Whenever I am checking if member is using AI tools for writing their posts I am checking all their post history, not just a single suspicious post.

If you are actually a quality poster, you will know eventually whether the thread is AI generated or not. Also, tracing the post history of the author is also a great idea if he’s really into AI majority of his posts, that way we can easily claim that he is not actually the author of his post. Furthermore, by using these AI detector tools, probably they can be reliable at some point but since they’re just computer generated as well, they can be prone to errors and mistakes. It’s still best to use your own intelligence and tell whether the post is AI generated or not.

[1715407052]

1715407052

[1715407052]

1715407052

[1715407052]

1715407052

[1715407052]

1715407052

[Viscore]

Whenever I am checking if member is using AI tools for writing their posts I am checking all their post history, not just a single suspicious post.

I think I said this before, but as a merit source it's disheartening to think it's come to that.  I've been mulling over my long-standing offer of post history reviews, but if I have to check multiple posts on a single member's profile just to see if they're written with AI or help from AI, that could turn into a full time job.

I've got to read this thread closely and see about which AI-detecting tool would best suit my needs, because I think I'm going to have to start using one eventually.  And Jesus, me being pretty dang tech-ignorant presents a challenge I'm presently too torpid to take on.  Foxpup, grant me strength!!

I guess that won’t be an easy job at all. Checking on post history takes a lot of time, but if you are really sincere on your goal, then you have to do it. However, that’s why we have this AI detector tools to help us determine if these posts are AI generated or not. But relying on them completely might also ruin your purpose as they can’t be reliable 100%. So you have to go through checking their previous posts and check also which of these tools will have higher chances to examine correctly.

[KingsDen]

Keep in mind that spambusters are probably catching just a fraction of those who use AI tools.

Seems you exaggerated here. You mean there are much more uncaught AI users here in the forum. What then is theymos waiting for to come up with rule regarding AI in the forum?

I don't want to automate anything related with AI, and don't fully trust any tool that is checking for AI generated posts, but using multiple tools is making those chances much higher.
Whenever I am checking if member is using AI tools for writing their posts I am checking all their post history, not just a single suspicious post.

If you are actually a quality poster, you will know eventually whether the thread is AI generated or not. Also, tracing the post history of the author is also a great idea if he’s really into AI majority of his posts, that way we can easily claim that he is not actually the author of his post. Furthermore, by using these AI detector tools, probably they can be reliable at some point but since they’re just computer generated as well, they can be prone to errors and mistakes. It’s still best to use your own intelligence and tell whether the post is AI generated or not.

Reading the whole post history of a suspect is a good idea to detect AI posters. The only problem with this is if the AI user is consistent in its usage. It will be difficult to detect. But if they use AI most times and the other times they make posts themselves, the discrepancy in their post will show that they use AI.

[Tungbulu]

I don't want to automate anything related with AI, and don't fully trust any tool that is checking for AI generated posts, but using multiple tools is making those chances much higher.
Whenever I am checking if member is using AI tools for writing their posts I am checking all their post history, not just a single suspicious post.

If you are actually a quality poster, you will know eventually whether the thread is AI generated or not. Also, tracing the post history of the author is also a great idea if he’s really into AI majority of his posts, that way we can easily claim that he is not actually the author of his post. Furthermore, by using these AI detector tools, probably they can be reliable at some point but since they’re just computer generated as well, they can be prone to errors and mistakes. It’s still best to use your own intelligence and tell whether the post is AI generated or not.

It's actually quite very easy to detect AI generated posts except maybe the author generated the post and used a paragrapher tool or manually paraphrased the content before posting, that way, it'll be quite difficult for AI content detectors to actually detect it was generated by an AI. Most of these spammers have actually grown pretty smart by the day, by learning from others' experiences, they manage to device mesns to cover their tracks lol.

[Ludmilla_rose1995]

Can we have a feature where the AI generated content automatically get flagged as AI generated content so the reputed members may not reply to that post/thread and just report it when they see that flag. I mean if it's possible to make a patch where users get information about AI generated content then they may not waste their precious time in replying to such posts/thread which aren't written by a human but are generated by bots. I need your valuable feedback and I also want to know that is it even possible to make a patch that could flag such content?

I'm sure it won't be that easy to find tools like this, that means AI versus AI, AI posts can only be detected with AI tools

for now all we can do is report posts that AI suspects are on topics that discuss that

[DabsPoorVersion]

I don't want to automate anything related with AI, and don't fully trust any tool that is checking for AI generated posts, but using multiple tools is making those chances much higher.
Whenever I am checking if member is using AI tools for writing their posts I am checking all their post history, not just a single suspicious post.

If you are actually a quality poster, you will know eventually whether the thread is AI generated or not. Also, tracing the post history of the author is also a great idea if he’s really into AI majority of his posts, that way we can easily claim that he is not actually the author of his post. Furthermore, by using these AI detector tools, probably they can be reliable at some point but since they’re just computer generated as well, they can be prone to errors and mistakes. It’s still best to use your own intelligence and tell whether the post is AI generated or not.

It's actually quite very easy to detect AI generated posts except maybe the author generated the post and used a paragrapher tool or manually paraphrased the content before posting, that way, it'll be quite difficult for AI content detectors to actually detect it was generated by an AI. Most of these spammers have actually grown pretty smart by the day, by learning from others' experiences, they manage to device means to cover their tracks lol.

Manually paraphrasing the content before posting still leaves a track that the content is AI-generated. It wouldn't be that hard to detect and if you think that the post you have seen is suspiciously AI-generated post, you can test it if it is really written by a human, or if the user has used an AI tool.

[Odohu]

In confirmation of my words above. Here are two examples. Two outrage threads would have appeared by already

Example 2

I have noticed an increased in the number of newbies using AI to make their posts, maybe they think that is the way to impress the older members of the forum. It is rather unfortunate that many of them are too lazy to at least read the forum rules that forbids such actions, hence the corresponding increase in newbie accounts being banned. I do not know what will be the solution but I think that any new member who cannot read the rules of the forum and who is not ready to learn and contribute to discussion base on his knowledge but resort to copying and pasting garbage from AI do not deserve to be here.

This is what I'm talking about.

I have never used AI before and I do not plan to do that soon even in my academic work (when I start my post graduate programs) as I find it a little lazy. With this result, I completely doubt the authenticity of those AI detectors and it simply means that anything written in fine and error-free English will be flagged as AI generated. I think we have to come to a consensus before a bad AI detecting tool causes chaos in the forum.

I tried checking how reliable the referenced AI detecting tool is with the following steps.
1. I posted my above comment in the took and got below results

https://sapling.ai/ai-content-detector/6fac85451e64588e260f1d647c36b87e

When I adjusted the statement by simply removing the word in the bracket  I got the result below:

https://sapling.ai/ai-content-detector/6fd3520673d3f484186519cf068065bb

I do not trust an AI detecting like this and it does not look good drawing conclusion from such a tool.

Below is the gptzero result of the same post flagged as AI- generated by sapling

[Tungbulu]

I don't want to automate anything related with AI, and don't fully trust any tool that is checking for AI generated posts, but using multiple tools is making those chances much higher.
Whenever I am checking if member is using AI tools for writing their posts I am checking all their post history, not just a single suspicious post.

If you are actually a quality poster, you will know eventually whether the thread is AI generated or not. Also, tracing the post history of the author is also a great idea if he’s really into AI majority of his posts, that way we can easily claim that he is not actually the author of his post. Furthermore, by using these AI detector tools, probably they can be reliable at some point but since they’re just computer generated as well, they can be prone to errors and mistakes. It’s still best to use your own intelligence and tell whether the post is AI generated or not.

It's actually quite very easy to detect AI generated posts except maybe the author generated the post and used a paragrapher tool or manually paraphrased the content before posting, that way, it'll be quite difficult for AI content detectors to actually detect it was generated by an AI. Most of these spammers have actually grown pretty smart by the day, by learning from others' experiences, they manage to device means to cover their tracks lol.

Manually paraphrasing the content before posting still leaves a track that the content is AI-generated. It wouldn't be that hard to detect and if you think that the post you have seen is suspiciously AI-generated post, you can test it if it is really written by a human, or if the user has used an AI tool.

If well edited and paraphrased, I doubt that a detector would be able to detect, it's programmed and can be easily confused if it's well edited by human. Ai detectors are programmed detect contents created by an AI, and AI isn't human, it can't just create new ideas, it surfs the internet and provides you with informations provided by other humans from other sources, so if it's not exactly the way it was copied, it'll be difficult to detect. It's just like inviting a  thief to detect and identify another thief... If that other thief dresses like a cop, he'll be difficult to detect.

[Casdinyard]

Let me just say, that while you're at it trying to find ways to figure out whether someone's plagiarizing off of ChatGPT or something, that you stray away from Copyleaks.ai. Like at this point every AI checker on the market's a better option than this one, let me tell you why.

Last year, I almost got laid off from copywriter job, cause one of their clients thought I was actually using ChatGPT to write my shit and then pass it off as my own work, which is a blatant lie that I had to fight and defend myself against. To prove myself, I literally took an excerpt from the bible, fed it to Copyleaks.ai, and voila, apparently even the writers from 1000 years ago (approximately) used ChatGPT to write their verses on the bible as well, who would've thought!

At the end of the day due to the massive embarrassment they allowed me to keep my job lol, and while I may have been very lucky to be able to defend myself against people who thought I was writing with AI assistance on me, a lot of the people who would be suspected (and are innocent) will not be, so I suggest that we make a mega-thread or a tribunal of some sorts for suspected AI-generated content, and from there, perhaps run their shit through 3 different AI checkers, and those who would get a red flag on 2 or more will be considered AI content and would be punished.

[nakamura12]

It isn't easy to detect an AI generated content unless it is very obvious and when you are using AI detection method that's why there are some reputed forum members who replied there. For now, we can only do it manually and then report it. There's nothing we can do to stop other people to use AI generated content to post here. What do you guys think when a forum rule is added which is using AI generated content will receive negative tag when proven?.

[FinneysTrueVision]

I have never used AI before and I do not plan to do that soon even in my academic work (when I start my post graduate programs) as I find it a little lazy. With this result, I completely doubt the authenticity of those AI detectors and it simply means that anything written in fine and error-free English will be flagged as AI generated. I think we have to come to a consensus before a bad AI detecting tool causes chaos in the forum.

Posts that are under 10,000 characters, which most posts on this forum are, will have a higher rate of false positive. In my experience, Sapling.ai has one of the highest rates of falsely saying something was generated with AI, that’s why I only use it as evidence in an accusation when at least 2 or more other AI detectors agree with it. In your case, I checked your post on three other detectors which appear highly in search results and the results were 0% probability on two of them and the third was only 5% probability of being AI-generated.

I also don’t get the sense from any of your recent posts that you use AI. Any reasonable person, looking into suspicious accounts, would quickly move on after seeing your post history.

Even though false positives exist, it is still worth investigating accounts that are actually spamming, as long as we look at a full body of evidence rather than just a single result from an unreliable tool.

[DabsPoorVersion]

Manually paraphrasing the content before posting still leaves a track that the content is AI-generated. It wouldn't be that hard to detect and if you think that the post you have seen is suspiciously AI-generated post, you can test it if it is really written by a human, or if the user has used an AI tool.

If well edited and paraphrased, I doubt that a detector would be able to detect, it's programmed and can be easily confused if it's well edited by human. Ai detectors are programmed detect contents created by an AI, and AI isn't human, it can't just create new ideas, it surfs the internet and provides you with informations provided by other humans from other sources, so if it's not exactly the way it was copied, it'll be difficult to detect. It's just like inviting a  thief to detect and identify another thief... If that other thief dresses like a cop, he'll be difficult to detect.

If you mean that well-edited and paraphrased is to completely edit the whole content, do we still consider it as AI-generated content? As I said, there will always be a trace and we can investigate the post history of that user to quality check the posts if there are traces of using AI-generated content.

In the end, it will be our responsibility to provide proof if we think that a certain user is using AI to produce the content of his post.

[Tungbulu]

Manually paraphrasing the content before posting still leaves a track that the content is AI-generated. It wouldn't be that hard to detect and if you think that the post you have seen is suspiciously AI-generated post, you can test it if it is really written by a human, or if the user has used an AI tool.

If well edited and paraphrased, I doubt that a detector would be able to detect, it's programmed and can be easily confused if it's well edited by human. Ai detectors are programmed detect contents created by an AI, and AI isn't human, it can't just create new ideas, it surfs the internet and provides you with informations provided by other humans from other sources, so if it's not exactly the way it was copied, it'll be difficult to detect. It's just like inviting a  thief to detect and identify another thief... If that other thief dresses like a cop, he'll be difficult to detect.

If you mean that well-edited and paraphrased is to completely edit the whole content, do we still consider it as AI-generated content?

Most definitely, any content that's not entirely written by human or the writing skill/pattern/idea wasn't initially yours but generated by AI or a website and altered by human and presented to be his own idea is considered to be plagiarism and fake.

Although, people can do researchs and gain knowledge online or from other sources and then come to share the knowledge, but it shouldn't be copy pasted, and there should be need to referencing when necessary.

[Lucius]

Nevertheless, this detector is on the list of recommended detectors in this topic. According to my observations more less adequate detector is contentatscale (not advertising).

I don't know why it is on the list (still), because @nutildah surely knows that this detector gives false positive results in most cases. Considering that this detector is the last one on the list, how did you come up with the idea of checking posts with it?

Basically, look at the post from @Odohu, which is just one more in a series of confirmations that only small changes can show completely different results.

[Albarq]

It's best if we do it ourselves. You can guess that the post was created using AI. This is a complicated problem and affects the forum for AI detection features to check and report a user may get away with it but soon it can be removed regardless of all posts it may be a problem in the future .

[Xal0lex]

Nevertheless, this detector is on the list of recommended detectors in this topic. According to my observations more less adequate detector is contentatscale (not advertising).

Considering that this detector is the last one on the list, how did you come up with the idea of checking posts with it?

I just took the list of recommended detectors from the aforementioned thread and checked two random posts with those detectors. To show that detectors can be wrong. That's exactly what happened.

[Doan9269]

It isn't easy to detect an AI generated content unless it is very obvious and when you are using AI detection method that's why there are some reputed forum members who replied there. For now, we can only do it manually and then report it. There's nothing we can do to stop other people to use AI generated content to post here. What do you guys think when a forum rule is added which is using AI generated content will receive negative tag when proven?.

What is only given them the privilege to continue with the use of AI in generating their post is because such is not against the rules and regulation of the forum, why have they stopped plagiarizing when they realized that such is the quickest means to get banned, if we can identify any post content that has AI generated post, with proven evidence using more than one AI detector, then such could be reported and members can take necessary action in giving them tags as it may applies.

i only see it that its for those who are lazy to make a constructive post from their head and present such will be the ones using AI to make post, this is only because they feels like they are been covered and wont get banned, unlike when it is direct plagiarism, but on a real sense, are they the actual author to those contents? are they not indirectly plagiarizing, the use of AI is nothing than a bad habit which we should all be frown at, we need genuine posters who will tell us something directly from how they feels and not stealing other peoples opinion and post.

[Sandra_hakeem]

As things stand, there will come a point where fighting the AI will be like trying to dry ice with a cloth... A never-ending fight, unfortunately

enhh, it's really pathetic!
I can remember writing on several topics, disposing the ideas of acquiescing AIs on the forum; AIs have done more harm already than it intends to do good in here... What's really disturbing is that so many people seem not to see anything wrong with it.

I just took the list of recommended detectors from the aforementioned thread and checked two random posts with those detectors. To show that detectors can be wrong. That's exactly what happened.

Detectors could be wrong too, For sure!! Nevertheless, it's always very obvious in some cases, even without them