Topic: Good Initiative by Royse777! - page 3. (Read 860 times)

IMO it's a good idea but the problem here is how to prove that the forum member uses AI tools upon posting.
A low-quality post considered made by AI?
Can AI possibly have wrong grammar when writing posts?

It might we can report low-quality posts but we can't rely on AI tools to determine or detect suspected AI posts.

Royse777's step to prevent Ai generated content on his campaign shows how much she cares about quality. Very much appreciated and unique approach of her to getting best result for the companies that hires her.
Though it is really tough to detect AI-generated content the reward for that is also good. Some members will accuse anyone without any kind of solid evidence, but Royse777 needs to verify the claim before banning anyone from his campaign and paying the reward to the accuser.
I can not agree with you here. I doubt that anyone here uses google translate for forum posting, and if he/she needs to translate to post he/she also needs a translator for reading the thread.

This is not a bad idea to report members who are suspected for using AI to generate their posts, but we should be really careful with this before making a report.
I did detect few members recently who probably did something like this and I reported them to moderators, I consider this to be spam that is similar like plagiarism.
Maybe it's time for unofficial forum rules to be updated so that members who are doing repeated AI generated posts can be banned.

I think I tested all of them that exist so far, some are doing very good detection job and others can be improved.
It's enough for me if I notice something suspicious in several posts from specific member, and if several AI detectors confirm my suspicion with high percentage.

OMG, that's just straight-up awful.  Glad it got reported--and yet I still wonder if that was from a bot/AI or just some moron who's ignorant as to what this forum is all about simply plagiarizing info about wallets in general (the spiffy leather kinds in particular, lol). 

That post brought me back to 2016 or so, when it seemed like there were tons of posts like that from account farmers just ranking up accounts based on activity alone.  I'm going to quickly forget ever reading that, and I will mention it no further.

Sometimes is way too easy because of completely misunderstanding the meaning of terms. If you want a good laugh, see here.
Sometimes they just write some more or less coherent (techno)babble. Keep in mind, those AI bots are still early.

Still, you're right, if a good AI is used, it may become difficult to catch them (sooner or later). And I expect that more and more of the long 0-value posts be reported as AI texts if they're written proper English (which may or may not be correct).


Indeed.

Yea its a great initiative by Royce777.
I think if the reporter of an offending post/poster doesnt get rewarded for an AI post
but the post is deemed spam the reward should still be awarded.

The idea by Royce777 not only activates all other campaign participants into
detective mode it also acts as a deterrent to those who would consider using AI.
Maybe it will raise the question in potential offenders, Is it worth possibly losing a
weekly wage and seeing it awarded to someone else?

It's not new to offer a reward to find bad users in a campaign, several campaign managers have done that. It's good business: if there's a bad apple in the campaign, he can be removed without additional cost and without much effort.
I've already found several AI chat bot spammers recently, but it's going to be hard to prove with 100% certainty.

It's impossible to be 100% sure, and all detectors are flawed. But as a human, it's quite easy to detect, especially if a shitposter suddenly has a long post in perfect English, includes background information that wasn't asked for, and writes much more words than necessary. Like this:

---

---

It's a good start though, if shitposters aren't paid for AI chat spam, they have one less reason to do it.

So are bots the same as AI scripts or what?  I'm assuming there's not a hell of a lot of difference in either case, but "AI" just sounds so much more nefarious.  And I'm guessing that this forum has had a problem with AI/bots for a long time now, though they've probably gotten harder to detect.

If they can be caught, and if what they write can be detected as garbage.  I'm not sure I'd be able to tell the difference between a script and just some random bounty spammer who can't write English to save his or her life. 

My, my how times have changed.

I've seen in the past more than once notes in the bounties that those finding cheaters could get paid, in a way or another. This is the same logic, since posting AI crap in signature campaigns is cheating, not just spam.
Of course, it's harder to catch that with proper proof, hence the bigger the pay.

I think though that just reporting the spam - AI generated or not - can have the same effect on long term (if people do report at sight) because mods can still remove those posts (hence the bounty/campaign participants won't get paid). Of course, then the reporters won't get paid. But we can't just have them all  

However, indeed, not a bad move for the campaign manager!

This is indeed a good initiativenkt just to find Ai post but also spam posts. This just proves that some forum members are lazy enough not to write their own post and decided to use AI generated to post. How accurate is it to determine if it really is an AI generated or written by a lersont?. I don't know id there's an AI that makes post that it may looks like it's written by a person. I can only say my opinion since I am not an expert at that.

Please, report low quality posts too. It's all well, and good collecting a reward for this, but how we get to the bottom of it is removing the content from the forum, and potentially punishing any users that are posting absolute crap. I know reporting is a thankless task, but we really do appreciate it. I know the end result might not be want you want, and there sometimes might be disagreement, and unfortunately a lot of users care a lot about their percentage going down, but I'd rather you get 50% of the reports you made handled, than none at all.

For the most part; these detectors are awful, and can't be relied on. There's better one's out there, but still they're too inconsistent. Currently, I'm trying to wrap my own head around it, since I suspect this will continue to be a problem. I know chatGPT are coming out with their own version, which I'd assume will be much better at detecting it. However, I don't really want to rely on that if I can. Some of the texts are pretty easy to decipher, others can be difficult. Straight up identical copies as they're output from the AI itself, can be identified with a decent amount of certainty though. It's when they're changed it makes it difficult or sometimes the AI can be confused what you've asked it, and generate complete gibberish, those are obviously pretty easy to identify. At the moment, it's considering multiple factors, piecing those together, and then ultimately deciding whether it's AI generated or not.

@tranthidung, I don't think (like most other members) that this initiative is bad, but as I showed in the example, I don't see that at this moment there is a precise method that can distinguish AI content from what is unique. If the members of that or any other campaign use various AI detectors and report their results to the manager, he will face a very big challenge to distinguish what is a real problem and what is a false detection.

However, this whole thing about AI and its influence on the forum is nothing new, we discussed it back in 2020, and as things stand, we will increasingly doubt whether we are having a discussion with real people or artificial intelligence.

Not exactly.
Because the way of writing is quite different. Even so, there may be some danger of this happening, but it will be little.

Proving that AI is used is very difficult, so there has to be very clear evidence for this to happen, not just what is done in a translator.

It is good initiative as you said but its not going to be an easy job, AI finding tools are ineffective cause with litt8text spinning and adding punctuations here and there will give opposite results so we need something from the Microsoft to say whether its created by ChatGPT or not, Gates is just doing some business.

The initiative of Royse777 is a higher level of plagiarism catching by finding likely AI-generated content and then checking more for text spinning, plagiarism. I think it is what Royse777 wanted with this mini campaign.

I believe if posters are lazy enough to use AI-generated contents, they don't have enough time to mind about text spinning to disguise that those contents are not from AI machines.

Bounty managers will use another AI an army of bounty hunters for participants to identify and catch AI posters. Genial.

But seriously, how to define this line between human posts and posts made by AI? In some cases, AI posts look more humane than living people. A sensitive situation may arise when a manager mistakenly withdraws an award or kicks him out of a campaign. It seems that soon we will hear a sea of whining.

In general, this is a good initiative. If there is a ban on the use of AI, then this will stop participants from dishonest posting. It is not necessary for the manager to check whether post AI was written or not, but this way he will filter out the crooks and let the community know that he doesn't approve of AI posters.

This is the part that will get those transactions in indentifying AI abusers.
Never underestimate the power of Bitcoin! A truely excellent oppertunity for those who themselves have tried it. I am sure this lucrative offer makes his signature campaign more robust as well as secure if he is able to achive some good results out of it.

What I also fear is that many non English members who use Google translate or any other might get banned. Do guys agree with it?

Although some posts may look like they are the work of AI at first glance, how will you prove that they are really the product of an AI Chat? I have tried 5-6 detectors and I am not satisfied with the results, because even very small changes in the entered text can give completely opposite results.

Here is the part of your post that one detector says is even 93% fake.


Source

Bounty manager I think will all have to add this rule to their campaigners. Sounds fair for all because its gonna be cheating if they just use the AI. Its pretty common these days that AI does a lot of things including writing articles and codes. So its not surprising if someone uses AI in the campaigns.


Its been encouraged to help mods finding plagiarism but with AI, it makes it difficult for them to identify. Such tools should be useful, I think Theymos can install something like this on Simple machines if is available.

Royse777, please accept my rays of respect for you. If all managers make such decisions, everyone who is going to participate in their companies will think a hundred times about whether to trust their posts to ChatGPT or not. In the end, the decision on whether a post was written by a human or a robot will be made by the manager, and accordingly, the participants in his subscription company should make even more efforts, since today's detectors do not always correctly detect even the text written by a person.
In any case, I see that ChatGPT can be discoverable, and Royse777's solution encourages everyone to not only write posts but also carefully read what others write. And this is very good behavior in order for the forum to receive live communication.