What's AI Written Post? - page 2. | Bitcointalksearch.org

pinggoki

sr. member

Activity: 1666

Merit: 426

Quote from: jokers10 on November 15, 2023, 02:38:41 AM

Quote from: pinggoki on November 15, 2023, 02:27:44 AM

~

AI generated text in fact is a sort of compilation of human written texts from Internet, so there's nothing truly non human in there. Grin

Users who use AI are not newbies only, different sorts of scammers and campaign abusers do the same, because they are not interested in making efforts for spamming or posting from multiple accounts.

Will be interesting to read about the results of your experiment.

Oh yeah, I forgot those fuckers who spam and abuse campaigns, the fact that they're using AI to supplement their abuse and keep up with their post goes to show that these people are stupid and have a limited brainpower because they can't keep up with their posts for their different accounts so they resort to using AI which to me is sad and pathetic, trying to get ahead of the competition but lacking the brains to keep up is weird to me. Don't expect too much on my experiment, it's just a simple one that I've been curious to do since last night.

jokers10

legendary

Activity: 1974

Merit: 3049

Quote from: pinggoki on November 15, 2023, 02:27:44 AM

Don't worry, that's my plan. I would post the results here for everyone to see. With the way that you're explaining it though, it seems that AI detection tools are either not yet that advanced to detect AI generated text or AI generated text are so good now that they appear human, that's what I am getting from your reply. Reputable members don't really need the AI help, they've been on this forum for a long time and it's most likely that the users that use AI for their posts are newbies that's trying to find a shortcut into making a quality post.

AI generated text in fact is a sort of compilation of human written texts from Internet, so there's nothing truly non human in there. Grin

Users who use AI are not newbies only, different sorts of scammers and campaign abusers do the same, because they are not interested in making efforts for spamming or posting from multiple accounts.

Will be interesting to read about the results of your experiment.

pinggoki

sr. member

Activity: 1666

Merit: 426

Quote from: jokers10 on November 15, 2023, 02:08:36 AM

Quote from: pinggoki on November 15, 2023, 01:55:13 AM

~

You can check multiple texts of some users you are sure are writing by themselves and look how often a false positive result will occur. You'll see that most users will have 0% of false positives, very few will have 1 per hundreds. So if there are several posts written in a short period are detected by several most accurate detectors it is highly unlikely that it can be false positive result.

Try to do it by yourself. When I try to detect my own texts with several detectors I always get a negative result, doesn't matter if the text is written in English or written in Russian and then automatically translated via online translators. The same for some very reputable forum members who are definitely writing by themselves.

Don't worry, that's my plan. I would post the results here for everyone to see. With the way that you're explaining it though, it seems that AI detection tools are either not yet that advanced to detect AI generated text or AI generated text are so good now that they appear human, that's what I am getting from your reply. Reputable members don't really need the AI help, they've been on this forum for a long time and it's most likely that the users that use AI for their posts are newbies that's trying to find a shortcut into making a quality post.

jokers10

legendary

Activity: 1974

Merit: 3049

Quote from: pinggoki on November 15, 2023, 01:55:13 AM

Well, if false positives and negatives exist then the possibility of people being wrongly reported will go up because some people will probably think that if it hits the 80 to 100 percentage marker on AI detection tools then it's going to be an AI? I know that you can still use about 2 or 3 tools to confirm but what if those 3 were generating false positive or negative results?

You can check multiple texts of some users you are sure are writing by themselves and look how often a false positive result will occur. You'll see that most users will have 0% of false positives, very few will have 1 per hundreds. So if there are several posts written in a short period are detected by several most accurate detectors it is highly unlikely that it can be false positive result.

Try to do it by yourself. When I try to detect my own texts with several detectors I always get a negative result, doesn't matter if the text is written in English or written in Russian and then automatically translated via online translators. The same for some very reputable forum members who are definitely writing by themselves.

pinggoki

sr. member

Activity: 1666

Merit: 426

Quote from: jokers10 on November 15, 2023, 01:33:03 AM

Quote from: pinggoki on November 15, 2023, 01:26:54 AM

~

When you detect something on something you can get a positive (yes, something is something) and negative (no, something is not something). But the accuracy of a detector it usually not 100%, there is a probability of false results. You should keep it in mind when you ask ChatGPT on if some text is written by it or by human being.

I will do another experiment on this one to see if ChatGPT can detect it's own generated texts, I've only done it in one device so I thought I was sure that it's an accurate detector that someone's used an AI on their text.

Quote from: jokers10 on November 15, 2023, 01:33:03 AM

Quote from: pinggoki on November 15, 2023, 01:26:54 AM

~

Besides that ChatGPT is not the only one AI program, so there could be texts written by some other chat bots.

I know that, I just didn't mentioned it in my first reply since I've only offered one solution that can work on AI generated text detection.

Quote from: Jawhead999 on November 15, 2023, 01:38:24 AM

Quote from: pinggoki on November 15, 2023, 12:48:43 AM

~

Yep, but it's not ethical to claim all of the AI generated text are yours.

AI only help to comply a history or get a general solution/answer, it's need human hands to make it better.

But the way that AI is learning, aren't they taught how a human writes or asks questions and giving answers so technically it's still human questions that are given it's just that the AI has compiled it to answer a specific question.

Quote from: Jawhead999 on November 15, 2023, 01:38:24 AM

Quote from: pinggoki on November 15, 2023, 01:26:54 AM

~

False positives means you're not using AI, but the result in detector site you're using AI..

While false negatives means you're using AI, but the result in detector site you're not using AI. This happen when you're modified or paraphrase the AI text by adding or remove punctuation etc.

Well, if false positives and negatives exist then the possibility of people being wrongly reported will go up because some people will probably think that if it hits the 80 to 100 percentage marker on AI detection tools then it's going to be an AI? I know that you can still use about 2 or 3 tools to confirm but what if those 3 were generating false positive or negative results?

Jawhead999

legendary

Activity: 1862

Merit: 1209

Quote from: pinggoki on November 15, 2023, 12:48:43 AM

It's contradictory for you to say that they shouldn't use it if there's no value to it because the response of AI does have value it's just that people don't like that there's no organic thought in it. I don't know about being sick of bots, it's a matter of perspective, if you're benefiting from bots in some way I don't think that you're going to get sick of them but if you're the one that's being inconvenienced by it then you're definitely getting sick of bots, no questions asked.

Yep, but it's not ethical to claim all of the AI generated text are yours.

AI only help to comply a history or get a general solution/answer, it's need human hands to make it better.

Quote from: pinggoki on November 15, 2023, 01:26:54 AM

What do you mean by false positives and false negatives? Can you provide me an example so I can understand what you're trying to say, I've only heard of false positives and negatives in a medical setting so I can't be so sure that it's going to be the same thing as this one. I'm not well versed in technology jargons and terms so I am curious about what you're talking about.

False positives means you're not using AI, but the result in detector site you're using AI..

While false negatives means you're using AI, but the result in detector site you're not using AI. This happen when you're modified or paraphrase the AI text by adding or remove punctuation etc.

jokers10

legendary

Activity: 1974

Merit: 3049

Quote from: pinggoki on November 15, 2023, 01:26:54 AM

Quote from: jokers10 on November 15, 2023, 01:00:46 AM

ChatGPT gives both false positives and false negatives, so it's not accurate enough to make final conclusions just relying on the data it can provide. Some natural text can occasionally look like written by AI, it happens very rare and usually you'll get some single false positive on many hundreds of someone's posts, but it can happen. And some obviously AI written texts are too short or too modified to be accurately determined as AI generated, it happens also.

What do you mean by false positives and false negatives? Can you provide me an example so I can understand what you're trying to say, I've only heard of false positives and negatives in a medical setting so I can't be so sure that it's going to be the same thing as this one. I'm not well versed in technology jargons and terms so I am curious about what you're talking about.

When you detect something on something you can get a positive (yes, something is something) and negative (no, something is not something). But the accuracy of a detector it usually not 100%, there is a probability of false results. You should keep it in mind when you ask ChatGPT on if some text is written by it or by human being.

Besides that ChatGPT is not the only one AI program, so there could be texts written by some other chat bots.

pinggoki

sr. member

Activity: 1666

Merit: 426

Quote from: jokers10 on November 15, 2023, 01:00:46 AM

Quote from: pinggoki on November 15, 2023, 12:48:43 AM

~

ChatGPT gives both false positives and false negatives, so it's not accurate enough to make final conclusions just relying on the data it can provide. Some natural text can occasionally look like written by AI, it happens very rare and usually you'll get some single false positive on many hundreds of someone's posts, but it can happen. And some obviously AI written texts are too short or too modified to be accurately determined as AI generated, it happens also.

What do you mean by false positives and false negatives? Can you provide me an example so I can understand what you're trying to say, I've only heard of false positives and negatives in a medical setting so I can't be so sure that it's going to be the same thing as this one. I'm not well versed in technology jargons and terms so I am curious about what you're talking about.

jokers10

legendary

Activity: 1974

Merit: 3049

Quote from: pinggoki on November 15, 2023, 12:48:43 AM

Also, if you really want to spot someone who's using ChatGPT for their posts, you can just copy paste their post and ask ChatGPT if they've written that post and they will definitely be able to identify if the user that posted that AI generated post is stupid enough to not change a thing or two about it.

ChatGPT gives both false positives and false negatives, so it's not accurate enough to make final conclusions just relying on the data it can provide. Some natural text can occasionally look like written by AI, it happens very rare and usually you'll get some single false positive on many hundreds of someone's posts, but it can happen. And some obviously AI written texts are too short or too modified to be accurately determined as AI generated, it happens also.

pinggoki

sr. member

Activity: 1666

Merit: 426

Quote from: PrivacyG on February 13, 2023, 08:38:35 AM

Think about it. Some body comes over to Bitcoin Talk and asks a very specific question about Bitcoin Wallets. Ten users reply, out of which eight are posts generated by AI. This will cause a lot of misinformation due to the imperfect accuracy of Chat GPT. Or, you start a debate and five users are arguing through Chat GPT. This would be an endless debate between nobody and therefore it is so unnecessary honestly.

It could be argued that if they're using the updated version of ChatGPT, they would probably be able to have a somewhat accurate information which is going to be difficult for someone to spot although you can still see some inconsistencies. Also, if you really want to spot someone who's using ChatGPT for their posts, you can just copy paste their post and ask ChatGPT if they've written that post and they will definitely be able to identify if the user that posted that AI generated post is stupid enough to not change a thing or two about it.

Quote from: PrivacyG on February 13, 2023, 08:38:35 AM

Please, do not even bother trying to use it on the Forum unless there is something actually valuable you are bringing to it. I think many of us are so sick of bots already.

It's contradictory for you to say that they shouldn't use it if there's no value to it because the response of AI does have value it's just that people don't like that there's no organic thought in it. I don't know about being sick of bots, it's a matter of perspective, if you're benefiting from bots in some way I don't think that you're going to get sick of them but if you're the one that's being inconvenienced by it then you're definitely getting sick of bots, no questions asked.

nutildah

legendary

Activity: 3010

Merit: 8114

Quote from: shahzadafzal on November 14, 2023, 02:28:46 PM

Well the reason why those posts are never deleted because you can't be 100% sure if the post is created by AI.

That's not exactly true:

Quote from: jokers10 on September 30, 2023, 04:24:32 AM

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

Quote from: ?? on ??

It is accurate to say that Daniel Ricciardo currently competes for McLaren and not AlphaTauri. For the 2021 and 2022 seasons, Ricciardo is under contract with McLaren as of my knowledge cutoff in September 2021. Future changes to a driver's team are speculative and subject to a variety of conditions, including team decisions and contractual commitments. Which team Ricciardo will eventually join is hard to foresee.

Most AI posters aren't dumb enough to include this text in their posts, which is a dead giveaway that its AI-generated, but every once in a while the infamous "knowledge cutoff date" makes an appearance. Besides, a post only has to be deemed as spam to be deleted, not AI.

Quote from: shahzadafzal on November 14, 2023, 02:28:46 PM

Also it's very easy to fool AI tools too, just add to your prompt "Create a post that's hard to tell if AI wrote it, also throw in a few spelling and grammar mistakes"

Sure, and that does happen, I'm already seeing it in the Gambling section. Or more commonly the poster will change capitalization and remove spaces to make the text look more non-AI. But the thing is, shitposters use AI because they are lazy, which means they will likely be too lazy to come up with such a prompt.

PrivacyG

legendary

Activity: 882

Merit: 1873

Crypto Swap Exchange

Quote from: shahzadafzal on November 14, 2023, 02:28:46 PM

Also it's very easy to fool AI tools too, just add to your prompt "Create a post that's hard to tell if AI wrote it, also throw in a few spelling and grammar mistakes"

It will be sad when this tool becomes so widely utilized and popular we will get waves and waves of Bitcoin Talk AI members who we will not be able to distinguish from real members.

Artificial Intelligence is useful for many reasons. But it is not useful at all on a Forum where people talk to each other. It is like these annoying Reddit bots. You will get their response to your topic and you will end up talking to a machine as if there was some body breathing and living for real behind a screen. Creepy. And sad.

Rruchi man

legendary

Activity: 1498

Merit: 1116

Top-tier crypto casino and sportsbook

Quote from: LDL on February 13, 2023, 04:49:01 AM

AI writing post against forum policy?

Using AI to write on forum removes the originality that a forum is meant to have, and if it permitted, it will make discussions in forum sound very inorganic. You don't have to be under the pressure of trying to impress when you write on the forum, because that is when you can be tempted to want to do extra and use AI. It is bad, and is not a pattern that should be encouraged.

Quote from: LDL on February 13, 2023, 04:49:01 AM

What are the pros and cons of using AI writing systems?

Since we are discussing AI, let me think out loud! what are the chances that if the same AI tool write on a topic asked by two users, what are the chances to provide the two users two different write ups on the same exact topic?

shahzadafzal

copper member

Activity: 1526

Merit: 2890

Quote from: lovesmayfamilis on November 14, 2023, 04:23:59 AM

I haven't seen the whole story, but it looks like you're right. The account is already blocked.
Another strange thing is that sometimes on the topic of the use of AI, we provide accurate data that accounts use AI,, but their posts are never deleted. However, if you look at your merit received today, the administrator is also against the AI. That is, we can conclude that it is forbidden to create posts using a robot, but as always, what happens with plagiarism, moderators make different decisions.

Well the reason why those posts are never deleted because you can't be 100% sure if the post is created by AI.

Yes I agree there, could be enough doubts to confirm if the post is written by AI but still it's not 100%. You are still not 100% certain that yes this post is written by AI. So this small benefit of the doubt goes in favor of the poster.

Also it's very easy to fool AI tools too, just add to your prompt "Create a post that's hard to tell if AI wrote it, also throw in a few spelling and grammar mistakes"

Edit: Am I giving ideas to AI posters?

KingsDen

legendary

Activity: 1316

Merit: 1089

Goodnight, o_e_l_e_o 🌹

Quote from: LDL on February 13, 2023, 04:49:01 AM

How to post on bitcoin forum using artificial intelligence system?

This question seems like, how to rob a bank. I might know but will not tell you because of the fear of the authority. Meanwhile, no one will teach you to cheat the system.

Quote from: LDL on February 13, 2023, 04:49:01 AM

And is anyone on this forum using the AI writing system?

If no one was using it, why would some campaign managers make rules against it?

Quote from: LDL on February 13, 2023, 04:49:01 AM

What are the pros and cons of using AI writing systems?

The pro is equal to the con.
Using AI to make a post will make you appear more intelligent than you are. This is a pro right? But on a second thought it wll make you dump. You might not a have a mind of yours to reason.
This is a discussion forum, so you should be able to flow with discussions naturally and not aiding yourself with an AI.

Quote from: LDL on February 13, 2023, 04:49:01 AM

AI writing post against forum policy?

Unofficially yes, it is against the forum policy.
Officially no, I haven't seen theymos say anything about AI posting.

itorai

member

Activity: 176

Merit: 34

Reward: 10M Shen (Approx. 5000 BNB) Bounty

Quote from: _act_ on February 13, 2023, 05:53:47 AM

Quote from: LDL on February 13, 2023, 04:49:01 AM

How to post on bitcoin forum using artificial intelligence system?

And is anyone on this forum using the AI writing system?

What are the pros and cons of using AI writing systems?

AI writing post against forum policy?

I do not know how to use AI to post automatically on bitcointalk, but I know people do against the use of bots to post because it may not give the accurate information needed and some information from bots can be spam. The best is to write posts offhand from what you know and from the experience that you have.

Yes, I agree with your opinion, indeed if you use the tool as mentioned it might be accurate, but it would be better if we discussed it humanely.

Agbe

legendary

Activity: 1106

Merit: 1372

Quote from: LDL on February 13, 2023, 04:49:01 AM

How to post on bitcoin forum using artificial intelligence system?

When someone asked a question in ChatGTP popularly known as AI and got the answer from there and copy everything and paste it here in the forum.

Quote from: LDL on February 13, 2023, 04:49:01 AM

And is anyone on this forum using the AI writing system?

I think many of them have been caught both in the gambling section and the other boards.

Quote from: LDL on February 13, 2023, 04:49:01 AM

What are the pros and cons of using AI writing systems?

It gives you straight answer without you thinking much, that is one of the advantages and when copied what AI have to you and paste it here as your original post then it becomes the disadvantage of using the chat.

Quote from: LDL on February 13, 2023, 04:49:01 AM

AI writing post against forum policy?

Yes because it also one of the clarification of plagiarism. When you copy AI work and paste it here as your original work then you have plagiarized another person work because ChatGTP does not have it own information but other people information so if you are talking information from there then you have to mention the source. And if you didn't provide the source of the information then it has violated the rules of the forum.

Learn Bitcoin

hero member

Activity: 462

Merit: 767

Instant cryptocurrency exchange with own reserves!

Quote from: lovesmayfamilis on November 14, 2023, 04:23:59 AM

However, if you look at your merit received today, the administrator is also against the AI. That is, we can conclude that it is forbidden to create posts using a robot

Unfortunately, No one could convince mprep, who wrote the Unofficial list of (official) Bitcointalk.org rules, guidelines, FAQ. We have seen good reports against AI-generated posts, yet forum moderators did not add such rules to the list of Bitcointalk.org rules. We can suggest that spammers not create AI-generated posts,, but there are no rules.

This is another reason we see brand new account just signed up and started posting using ChatGPT or other AI. These spammers should be banned. But they are getting free pass even after the positive report because there is no rules for AI posting yet.

FatFork

legendary

Activity: 1820

Merit: 2700

Crypto Swap Exchange

Quote from: lovesmayfamilis on November 14, 2023, 04:23:59 AM

Quote from: jokers10 on November 13, 2023, 10:31:55 PM

One day it will lead you to a ban or getting a tag. Your other post is 100% detected by all detectors as AI generated...

I haven't seen the whole story, but it looks like you're right. The account is already blocked.
Another strange thing is that sometimes on the topic of the use of AI, we provide accurate data that accounts use AI,, but their posts are never deleted. However, if you look at your merit received today, the administrator is also against the AI. That is, we can conclude that it is forbidden to create posts using a robot, but as always, what happens with plagiarism, moderators make different decisions.

The account got banned (nuked) because everything he wrote up to that point seemed to be made by AI. So it was pretty obvious to the mods that he only set up the account to spam junk: https://ninjastic.space/search?author=NelonExer

But it's not always so cut and dried and sometimes it's probably tricky to judge if something is fully AI-generated or just uses AI tools like machine translation or spellcheck and grammar fixes.

lovesmayfamilis

legendary

Activity: 2072

Merit: 4265

✿♥‿♥✿

Quote from: jokers10 on November 13, 2023, 10:31:55 PM

One day it will lead you to a ban or getting a tag. Your other post is 100% detected by all detectors as AI generated...

I haven't seen the whole story, but it looks like you're right. The account is already blocked.
Another strange thing is that sometimes on the topic of the use of AI, we provide accurate data that accounts use AI,, but their posts are never deleted. However, if you look at your merit received today, the administrator is also against the AI. That is, we can conclude that it is forbidden to create posts using a robot, but as always, what happens with plagiarism, moderators make different decisions.

Topic: What's AI Written Post? - page 2. (Read 937 times)