Jump to content

Recommended Posts

Posted

OK, here's a solution. 

 

1. Those who want to post screen shots of ''accurate'' (or inaccurate) AI content detectors should do it. 

2. Those who believe the detectors are not producing accurate results should not take them serious!   

 

How about I throw a party with zero soft drink and zero decency? 

  • Like 8
  • Up 2
Posted (edited)
3 minutes ago, maitasun said:

Two can play this game, UK...

So you want to continue letting people who we know for a fact are not even able to spell Fiverr correctly to post AI content as if it had been written by them, although WE ALL KNOW they're not able to do so, and falsely lead others to think that content is theirs?

This is a very sad joke you're defending.

You can just just say you believe it is AI content or show the screenshot of the detector with a disclaimer saying it's not always accurate (because it is true that it is not always accurate).

Or you can just flag the post with the reason being that you "believe it is AI generated content posted without citation (against rule 8 of the forum rules)" and it will be removed when the people who check those flags see it if they agree.

Edited by uk1000
  • Like 7
  • Up 3
Posted (edited)

People. I'm one of the users always calling out meksells using chatgpt. Just check the forum, and you'll see several examples. I'm obviously against it.

And I'm also against using "AI detectors", because they are crap. They don't work. They can't work, by design. 

You don't need "AI detectors" to call out meksells using AI to write - it's obvious to a human anyway. I've called out multiple users for chat gpt answers, and I never used a detector. It's not needed.

And you can't use "AI detectors" to catch sophisticated users who actually know how to use AI without getting caught. They won't work, at a certain point it's impossible to tell "real AI text with some work done on it" from a false positive.

So it's a moot point.

Edited by visualstudios
  • Like 10
  • Up 3
Posted
8 minutes ago, maitasun said:

So you want to continue letting people who we know for a fact are not even able to spell Fiverr correctly to post AI content as if it had been written by them, although WE ALL KNOW they're not able to do so, and falsely lead others to think that content is theirs?

 

4 minutes ago, visualstudios said:

You don't need them to call out obvious meksells using AI to write - it's obvious to a human anyway.

 

SoTrue.jpeg.09ba37bea1393d971eec9adf6739915b.jpeg

  • Like 7
  • Up 1
Posted

It's so obvious that content generated through AI specifically chat gpt has a same pattern and language and can easily be detected without using any detector. However, as pointed out by @uk1000, how can it be detected if, let's say, I post someone else's sophisticated article and ask chat gpt to write it a technical way. Then I copy that rewritten article and ask bard to rewrite it using tough vocabulary and change the style and pattern of the article?

  • Like 9
Posted (edited)

At a certain point, it's impossible to say if a text is human or AI. Where's that point?

1 - A sufficiently advanced AI will be able to mimic a human perfectly. That means being indistinguishable from one, therefore undetectable. We're not quite there yet, but will be soon.

2 - A sufficiently advanced user will be able to tweak an imperfect AI output in a way that it will be undetectable, by injecting "humanity" into the output. We're perfectly past this point now.

Obvious AI content, used by clueless users, will be obvious until we get to 1). We don't need detectors for that, it's obvious. 

Once we get to 1), or while we're dealing with 2), no detector will be useful.

Edited by visualstudios
  • Like 11
Posted (edited)
1 hour ago, vickiespencer said:

What.thumb.png.610a14503dafda1d636848766026a232.png

I meant "screenshots of AI content detectors" not "screenshots of AI content generators". I fixed that post 😬

Edited by uk1000
  • Like 6
  • Thanks 2
Posted (edited)
3 hours ago, visualstudios said:

it's obvious to a human anyway.

This was my original point. We're talking about people who are scamming by their very method of posting, they are cheating to "rank gig", so why all the commotion if we add a silly AI detector to tell us what we already know?  

Edited by newsmike
  • Like 8
Posted (edited)
1 hour ago, newsmike said:

so why all the commotion if we add a silly AI detector to tell us what we already know?  

No commotion for me, I just don't think that's the point - I've always called the chat gpt copy-pasters, no need for detectors. My point is that legitimising those detectors, that are anything but trustworthy, can be dangerous. If you feed them the posts of users I'm fairly certain are not using AI, including mine and yours, they'll return false positives half the time. So it's counter productive to call on them as proof of anything, as it can backfire. And there's no need to anyway, we all can see when some user on the forum is using chat gpt.

Now, using them as a joke is fine, of course, but not everyone gets the joke, and I can just see the day when someone will accuse me of being a bot, with "proof", and I don't like it. Or, god forbid, the platform itself decides to introduce some form of AI detection (which is an AI itself, and they do love to use AI to automate decisions, with very questionable results) and start flagging gigs, sellers, etc. for "AI generated content". That's very dangerous.

For now, I think that the only way I trust to decide on if something/someone is human, is having a (qualified) human analyse / interact with it. Leaving that to a machine doesn't sit well with me.

Edited by visualstudios
  • Like 9
  • 4 months later...
Posted

Hey Fiverr Community!

Creativity knows no bounds, and it's time to unleash our imaginative spirits! Let's dive into a world of unique possibilities. 🚀

Question:
What's the most creative gig you've ever come across or offered on Fiverr? 💡 Share your extraordinary experiences! Let this be an inspiration for all of us to think outside the box and discover the limitless potential in our talents. Your story might just spark the next big idea! 🌟

Join the conversation and let's celebrate the vibrant creativity within our community. Share those gems that left you in awe or the gig you crafted that stands out from the rest. Excited to see the magic unfold! 💼

#FiverrCreativity #UniqueGigs #InspireInnovation

  • Like 13
  • Sad 1
Posted
20 minutes ago, startalent2024 said:

let's celebrate the vibrant creativity within our community.

 Thank you, ChatGPT, for writing such an impressive unique post for us. 😶

  • Like 14
  • Up 1
Posted
16 minutes ago, oymusafir said:

 Thank you, ChatGPT, for writing such an impressive unique post for us. 😶

Thank you for sharing your thoughts. Tools like chatGpt came to make life easier for us. They help me a lot and they make me more creative. However, the main idea was mine and of course GPT made this unique and nice text.

  • Like 7
  • Sad 1
Posted
2 hours ago, startalent2024 said:

Thank you for sharing your thoughts. Tools like chatGpt came to make life easier for us. They help me a lot and they make me more creative. However, the main idea was mine and of course GPT made this unique and nice text.

They make you dumber, not more creative. It's very disgusting to read ChatGPT text for everyone involved in seeing it.

  • Like 8
  • Up 2
Posted
9 hours ago, spintele said:

They make you dumber, not more creative. It's very disgusting to read ChatGPT text for everyone involved in seeing it.

It seems you did not use ChatGPT or similar tools before. Nevertheless, maybe you were right about the text. Maybe it was better to keep it shorter to be easier to read. But using words like dumper was not suitable in the forum.

  • Like 8
Posted
10 hours ago, startalent2024 said:

It seems you did not use ChatGPT or similar tools before.

You are violating Forum rule number 8. On the forum, you can't post AI-generated content.

  • Like 8
  • Up 1
Posted
1 hour ago, oymusafir said:

You are violating Forum rule number 8. On the forum, you can't post AI-generated content.

Actually, it wasn't completely generated by AI. The original version was more than 20 lines. I made many changes on it, I just used the tools to make better text. However, thank you for reminding me, I didn't know that.

I think people are here to fight each other, but I thought the purpose of the forum was to learn and help each other to make a better selling experience on Fiverr.

  • Like 5
Posted (edited)
1 hour ago, startalent2024 said:

I think people are here to fight each other, but I thought the purpose of the forum was to learn and help each other to make a better selling experience on Fiverr.

People come here to talk to other human beings, not robots. I believe many find it rude when you speak to them through ChatGPT instead of using your own words. It makes it feel very impersonal.

Just imagine if everybody in this thread gave you ChatGPT replies. It would be robots talking to each other.

Would you be happy with that?

Edited by vibronx
  • Like 10
  • Up 1
Posted
26 minutes ago, vibronx said:

People come here to talk to other human beings, not robots. I believe many find it rude when you speak to them through ChatGPT instead of using your own words. It makes it feel very impersonal.

Just imagine if everybody in this thread gave you ChatGPT replies. It would be robots talking to each other.

Would you be happy with that?

Thank you for sharing your ideas. It seems I finally found someone with good intent who wants to teach something.

Yes, you are right, I am a new seller and it is my 4th post in the forum. I agree with you and I will consider it in the future.

 

  • Like 9
Posted (edited)
3 hours ago, oymusafir said:

On the forum, you can't post AI-generated content.

Yes you can as long as the source is cited. See Fiverr Forum rule 8 below:

Quote

Posting original material is encouraged. If you post information taken from the Fiverr site or other places, cite its source. This includes AI-generated content.

Edited by uk1000
  • Like 7
Posted (edited)
On 12/27/2023 at 9:24 AM, startalent2024 said:

However, the main idea was mine and of course GPT made this unique and nice text.

Yes, about 70% was your words. However, this part was not: 

Screenshot2023-12-28at12_53_13PM.png.124b64e020a9223d9c2244d423433982.png

 

As @vibronx pointed out, people come here to talk to other people not to interact with robots. If you stay around the Forum for long you will see that AI-generated posts are quickly removed. 

 

58 minutes ago, uk1000 said:

Yes you can as long as the source is cited. See Fiverr Forum rule 8

So, the next time @startalent2024, you want to use AI-generated text, please tell the reader that is what you are doing. However, the best thing to do is use your own words. AI-generated wording is too flowery and unnatural.  

Edited by vickiespencer
  • Like 9
Posted
1 hour ago, uk1000 said:

Yes you can as long as the source is cited. See Fiverr Forum rule 8 below:

Thank you for checking the rule and giving us the right information.
However, as I mentioned before, I don't use AI in the future to make my content better in the forum to respect others and the rules.

 

  • Like 5
  • Up 3
Posted (edited)
4 minutes ago, startalent2024 said:

However, as I mentioned before, I don't use AI in the future to make my content better in the forum to respect others and the rules.

I have yet to look at your profile or gigs, but Fiverr's TOS says that if you use AI-generated text on your profile, you must cite the source, too. 

Edited by vickiespencer
  • Like 9
Posted
12 minutes ago, vickiespencer said:

AI-generated wording is too flowery and unnatural.

Surely if AI content is trained on millions of examples of actual human written text, and it learns and is designed to output the next token/word that it expects to see (based on all the training data it's seen, then surely it would be roughly the same type of wording as all the real text it's seen in the training data (eg. millions or more of real sentences). Different AI model will be trained with different data though but it's roughly how it will be, and the output will depend how they've got it set up (eg. if it's basing it on the most probable x amount of words each time - or how much randomness (eg. use of less common words)/uniqueness someone wants).

  • Like 6
Posted (edited)
2 minutes ago, uk1000 said:

Surely if AI content is trained on millions of examples of actual human written text . . .

I am referring to the AI-generated text that many of the Forum users use. It is very apparent that they are using it as it does not sound natural compared to their "normal" posts.  

Besides, what I wrote is my opinion of the way AI text sounds. It is not a fact. 

Edited by vickiespencer
  • Like 9

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×
×
  • Create New...