Bitcoin Forum

I stumbled on a thread created by @nutildah (https://bitcointalk.org/index.php?topic=5448111.0) and I was shocked and sad at the same time as this might be calling for a low-quality posting which I'm not ready for.

I believe we as humans should always strive to be a better version of ourselves, English improvement is not an exception, and I don't think BTT should take it away from us. I've been improving my English for years as an editor and book writer, and I've participated in many strict fora, of which one of them takes polished English writing as a requisite for post-approval before I joined BTT last year. They also recommend "Grammarly" for English improvement.

"Grammarly," an English software recommended by even schools has always been my English tutor and guide for the past 3 years, which has tremendously helped me, and of now, it hardly corrects 2-4 errors in what I write (including punctuation marks) because I was strict with its tutoring. Should I now retrogress my English writing because of BTT?

My average score on computer-based English tests is now 94%, so the forum should know that there 'must' be some exceptional writers. Some people coded the AI's English, and it's not from the heavens. So, it will be unfair to tag the post of a few exceptional English writers on the forum AI-generated when they are 100% innocent.

I was sad with the results seen when I decided to use two of the links (https://writer.com/ai-content-detector & https://x.writefull.com/gpt-detector) posted by @nutildah to test two of my written posts. The first post tested 100% and 99% human respectively. But the second post (https://bitcointalk.org/index.php?topic=5448086.msg62104267#msg62104267) tested 47% and 23% human respectively. This is absurd and unjust!

Personally, I love writing, I possibly don't know why I would want an AI to write a piece of cake for me. But thankfully, some people kicked against the idea, yet I want this to be a case study for the people concerned in the forum.

Peace!

You missed one crucial key points on that post:


One indicator said by nutildah is if the improvement happened in one night from a messy English to a flawless ones. I didn't look at your post history, but if I may draw a conclusion from your post above, your main language is not English and you're gradually improving your skills; vocabularies, grammars, phrases, slangs, etc. On that case, I don't see why you should feel shock or saddened with the thread, given it barely matched your criteria, nor that I find any part on that thread that encourage people, implicitly or explicitly, to deliberately "retrogress" their english writing on BTT just to avoid being detected as AI.

The forum always encourages improvement. There are literally threads filled with suggestion for non-english speakers to use a translator and then to double check the grammar with grammarly or other similar software to let them post something that's close to a perfect english. So thinking that the forum would prefer you to "downgrade" your english is somewhat baseless.

You also missed this post somehow:


The general consensus is we can't report posts for simply being AI; they have to be spam first and foremost. As far as trust ratings are concerned, spammers should only be neutral tagged at most.


Whats absurd is assuming any negative action would be taken against you for this result. Let's look at the results for your post from 5 detectors (its not long enough to be used by 2 of the 7 I listed here (https://bitcointalk.org/index.php?topic=5448111.msg62060187#msg62060187)):

Writer.com: 47% human
gptzero.me: "Your text is likely to be written entirely by a human"
contentscale.ai: 100% - Highly likely to be human
writefull.com: 77% AI
paraphrasingtool.ai: "It's likely that a human wrote this text"

It would therefore not be rational to conclude that your post was written by AI.

I believe anyone would be able to tell between user whso gradually improves their English over the years and one which suddenly moves from below average to perfect use of English. Self improvement is not being taken away from you.

It would be ironic for one to rely on a machine to detect machine generated text. You should not feel hard done if a machine adjudged your post to not be 100% human generated, it's a tool and could have flaws at times.

---

You can drop your reply in that thread since the discussion is ongoing

I am sure we already have and will have even more shitposters who use AI-generated content in the future. I don't like seeing it, and it's plagiarism. Cheap merits and ranking up are the reasons people will use it. Thus, it's cheating. It's not always going to be easy to differentiate someone with excellent English skills and a person who uses AI bots. A sudden change in the posting style is a dead giveaway though. Until they figure out that they better create brand-new accounts that post post only AI-content. But even then, they will make mistakes sooner or later.       

I don't think using AI generator is plagiarism, but if the AI generator took someone text, then it's good since some people can report it to make these shitposters get banned.

I think any user especially merit source need to give merit to a post which have an emotion and a subject. Usually AI generated post are objective and only discuss about the particular matter. Although there are few users are tend to explain particular thing in very objective.

Of course it's plagiarism. Plagiarism tools might not detect or classify it as such, but AI-generated content fulfills both conditions to be considered plagiarized.

1. You are using text that doesn't belong to you and which you didn't write. A bot wrote it.
2. You aren't posting a source from where or what you got the information from.

That's why it's plagiarism.   

I think it would be more classifiable to be plagiarism than even spamming or shit posting because some of them are mostly talking on point. A BoT generated work isn’t yours and that is what plagiarism is all about.

It is hard to detect this accounts when distributing merit because the merits source might just focus on the content shared and how it was of help rather than having to dig into the post to self if it is Al generated. Also most emotional posts are just merit fishing posts or even fabricated lies.

actually I disagree I do James Joyce stream of conciseness  on purpose and I have done it for years.

---

---

Actually I disagree with your idea that rapid improvement means a person is cheating with AI. This is my corrected version of the sentence above.  I know I did improper grammar in the first example. I know that stream of consciousness was misspelled.


I simply ask you if I did a google search and found this below

https://astrologicalpsychology.org/james-joyce-stream-of-consciousness/

Which allowed me to spell it correctly did I cheat?


  Do I give it a credit for showing me how to spell stream of consciousness correctly. Not needed in the first sentence as I did not get proper spelling. I would argue any correction made via a google search or a spell check simply does not need to be listed as a source.

 So if I write all of the above and ask an AI to polish it up am I cheating?

Which btw is why I did a deliberate choice to write stream of Consciousness style when I blog.


 Please compare the first sentence with thousands of my posts and you will read do that style often.

The last part of this post was stream of Consciousness :

 I do not spend the time to check all spelling and grammar as I want to get the idea out as it unfolds in my head in real-time.

Some time ago I had a discussion with @Lucius regarding people who have abused AI for certain purposes, especially in forums. Their silliness started when chatAI acted as a source of quality improvement, but it turned out to have a negative impact on the authenticity of the posters in the forum. I agree that the people behind the skills who use AI must be eradicated.

If you want to look smart and proficient but not from the original results of your own thoughts or writing, in the end, that person will be left behind by those who are still pouring out their content and ideas so what's wrong with never stopping learning?

I think the post quality of ChatGPT is decent and above normal because it's AI generated that polish the grammar of the info available pn the internet. I think the main issue here for ChatGPT post was the inability to have an organic discussion since this kind of post is suitable as a bot response and not with human emotion.

The danger of this tool is to kill the genuine discussion. Worst you are replying to someone that is just getting the answer on the internet without original input to the post.

Unjust? What's the injustice here?
A text analyzing labeled your text as xx% and xx% that, there is no "injustice" , you haven't been affected by their decision, you haven't suffered any damage, nobody cares what a text analyzer thinks or what the results are and I don't see why you would either!

But I can tell you for sure why some will get their posts marked in such a way, it's mainly because they try so much to write down a post like it's fucking essay that you're getting a mark on every single phrase that they end up with one huge wall of text that spins the same idea which stops looking like being normal text written by human and more like some machine. Let's not even mention the 3-4 liens quota, some simply force themselves to make that number of words characters, and of course when you try to stretch a 4 words phrase into 100 it it stops sounding human.

Oh and another thing, seems like all those analyzers dislike a text made out a ton of short phrases, don't know if that's something that ChatGBT does but this might get you a clue on why the high score.

And now going to check the above shitpost of mine with that detectors and:

Where is my cookie?

Sometimes I always check the results of the text that I make on the AI detection engine and the results are not very satisfactory and have tried several detectors but the results are always different from each other, meaning that there are still many deficiencies that occur and cannot be guaranteed to be 100% accurate, but I'm not going to be ironic because if you or you guys don't use Chat AI to post it will be fine, even if it's just shitposting but Chat AI is more dangerous because I will consider it plagiarism not your own work.

I wasn't sure how the detector works but I tried some of my posts and it's mostly 89% up human-generated, I agree that someone who has a good English post should not be punished just because he's good at English or have some kind of high-quality post. But when I think about AI it's really not that easy to prove that a post is an AI-generated text unless it was completely copied, so even someone that has good English for sure can argue when they get a low percentage on AI detectors.

For sure its rare to see a human-generated post getting a high score on AI detection.


As mentioned here, if there are changes to a member posting style and most of the posts have high chances of AI detection that could easily be a red flag, but still, there are for sure going to be processed on proving and defending if it's true or not.

OP, you could write your opinion in the thread you mentioned, which was created by nutildah.
I do not see much difference between the discussion there and what you say in your topic. I think that texts written by AI sometimes signal very strongly that the absence of various kinds of emotions can just determine that this is the work of a robot. Using Grammarly only corrects the speech that the person wrote. I think that's a big difference.

I envy the way you express your views in a clear but concise form @stompix. But people mostly from non-English speaking nations sometime can't express themselves accurately. So they feel one of the ways to further clarify their point to make the reader understand is to repeat their intentions using other words to avoid misconception. I am guilty of this shortcoming. Some writers want to express themselves comprehensively. They want to write it the way it flows from their mind, which is why their post most times are long. I am also guilty of this offense. I will have to improve my expression in the English language and also consider the length of my post because I will never want to be called a bot.

expressing oneself doesn't really mean writing some bunch of words to convince the reader, there are few words which you could just post that carry heavy meaning at same times summarized all you wanted to say.
When they said of spamming, it doesn't mean writing a very short words that results to spamming, you may write a very long and lengthy words but doesn't carry meaning would still results to spamming.

It is not good to accuse the forum of encouraging writing in bad English!!!

On the contrary, this effort on the part of the forum to combat posts written by artificial intelligence is evidence that the forum encourages good human writing and not the other way around!!

As for categorizing your writing as written by AI it should make you happy if you are innocent of this charge!!! Everyone should be grateful to the members who are trying to protect the forum from AI.

There are even times in which you can just directly construct the point of your reply using short sentences only as long as your message is clear, concise and it is readable you are already good to go. I am well aware that there are several members here that has very deep choice of words or their vocabularies, and it is not fair if we consider and judge them as bots. As for me personally, the meaning of "constructive post" doesn't mean how many characters you use or words and even grammars, it is how you convey your point of view since this forum isn't created for constructive criticism alone but also to provoke discussion and spread knowledge to one another.

Simple.

You don't have to write your posts with perfect English in Bitcointalk. We don't have An English as a Second Language (ESL) teachers or professors here (or do we have ?).

What do you need to communicate with other forum members?

• Write your posts succinctly enough to express your opinion shortly enough. Why short? Because if you write lengthy posts, less readers will feel comfortable to read it all. In addition, writing to lengthy posts, you can lose your main points in writing.
• Try to learn to write as best as possible individually in grammar as it will help you express your opinion better.

That's it. You don't have to rely too much on Grammarly or ChatGPT to compose your posts and get troubles.  :D

What matters in every written content is it's value not the grammatical syntax, rhymes, idioms etc. Those are for entertainment purposes to keep the reader, but if a content is valuable to the reader he doesn't care about the rhymes anymore, he focuses on the information he wants to get from the piece of writing. Most of the top writers in the past who made contents that changed the world and passed relevant information, check the contents now with the said grammarly, you'll see lots of errors; passive words, bad syntax, etc. Yet the contents are valuable due to the secrets and deep knowledge readers can grasp from reading them. Grammarly shouldn't be a tool for a good writer, he doesn't need it to make good articles, because grammarly makes mistakes too, it's a bot as well. So, in the forum if you are using grammarly it should be for typo purposes not to make your article look good or changing of your sentences, as the previous sentence you made yourself can be of good readability to the reader than the one grammarly suggests. What matters is the message you are about to convey to the forum, and I don't think chatGBT can be able to craft a valuable content which has not be posted anywhere on the internet, it has no brain and can't be compared to the brain of a new born baby.

Due to this inconsistencies in the above results, theymos is finding it difficult to make official rules about AI writing in the forum. I also think that some of those use the grammatical syntax of articles to determine if it is humanly written or AI written. But they ought to use the level of connectivity to an originating topic. Where this would be a problem to detect is when the AI generated post is actually the original post.

Op, you don't need to be discouraged and it does not call for retogration. The only implication of it is that it will drag the attention of campaign managers to your post especially Royse777 who already has rules against AI.

ChatGPT also has the capacity to be dumb:

https://i.imgur.com/wX9SPUy.png

I think OpenAI just trains ChatGPT to reject requests for it to write scripts that interact with any website and that don't have any APIs defined for it, and cites the reason of the rejection as "ToS violation". Which I think is lame. Why don't they just train it to write: "I don't know how to interface with this particular website"?

Nobody is going to ban you just for having great English. That's probably just one of the criteria they're suggesting you look out for after their style suddenly improves from going from broken English to flawless or near-perfect.

I really think theymos is going to have to both put a warning when new users sign up and also mass PM everyone warning them not to use AI posting and anyone doing so will be instantly banned like we do with plagiarism because it's going to get out of control. Using chatgpt to post is just as bad if not worse as plagiarism and it's probably harder to detect. I've long been calling for theymos to put a welcome message with some of the forum rules and links for further reading etc when people sign up and now I hope that spurs him on to do that or something similar.

I am not familiar with the concept, but I can figure it out based on this sentence of yours I quoted. You overdid it a bit, though. I don't remember seeing such posts from you. At least, not to that extent.
Nevertheless, you don't qualify as a shitposter, stream of consciousness or not.

Of course not. Google suggests corrections based on its dictionary anyway.

It's a delicate matter. I would say both yes and no. First of all, a shitposter doesn't know about stream of consciousness or James Joyce. Your average Bitcointalk shitposter doesn't know much about bitcoin or crypto either. The fact that you know and can switch between the two styles as desired removes all doubts. Doubts that didn't even exist.

Not every use of AI is immediately plagiarism. I see no problem in having an AI, Google Translate, Grammarly, or something similar correct the mistakes YOU WROTE. It's a problem if you ask the AI to create something for you because you can't write it yourself and then you present that and take credit for it. 

There's this common error we often make whenever we are typing in other to make a post, which is typographical error, this is very common and we all need to always proofread any post we are making before sending them, another thing is that learning never ends, we can all improves ourselves in many ways, there are instructional learning materials we can get in shops and also read online to aid our English speaking ability and dictions, but we must always reme that to err is human.

Usually applying those words without the reader comprehending what the writer is saying can like term to be spam in my own point of view, because we are meant to keep here clean with a plain English where it could be easily understand by the readers, and not just using all manner of words just for self expression.


This is good idea sir, maybe to reduce the stress and also help in keeping this forum clean from people that are too lazy to write  and creates their personal post, and contributes to the progress of this wonderful platform.

Just to add to that, AI generated posts is by all means a spam post in the regards that, your posting not because you want to but, because you are obligated to and source for assistance from some bot that might make posts that don't conform to the thread or perhaps it is a thread of its own but is of no interest to you that is making the post. The obligation just makes to either yourself or some attachment just makes you put it out there and that's not some good practice.


Haven't said that, the flaws in the system as alarmed by OP clears the air that these are not actual standards that is expected to work everytime but some speculative too to aid, in deciphering what could be from what ought to be. Not many users would be penalised out of this @OP.

Your conjugation and articulation @OP is cool and far from what is expected of an AI in the context of your subject which tells that, your use of English is good none the less.

Really? But I stand to be corrected if I'm wrong. So how about bloggers who pay content writers on Fiverr to write a full blog post to be posted on their news blog. Will you consider that as plagiarism? Because they actually didn't write the content, but rather paid someone to do that for them?? (i.e following your definition of plagiarism)

But I think for the fact that I had no source and can't be found anywhere on the web online is a reason it ought to have been non plagiarized content, because I'm kind of curious to have someone break this down.

Because I think the difference between plagiarized content and non-plagiarized content is that while plagiarized content has an already existing online source, non-plagiarized content has non.

Or do we have any contrary difference about this???

Members of the Bitcointalk forum should strive to improve their language skills. I believe AI tools can be helpful in identifying and correcting errors, they should not be relied upon as the sole means of improving one's language proficiency.

It is important for members to actively seek out opportunities to improve their language skills through reading, writing, and engaging in discussions with others. Additionally, forums like Bitcointalk can provide a supportive community where members can receive constructive feedback and guidance on how to improve their writing. Encouraging members to take an active role in improving their language skills can not only enhance the quality of discourse on the forum, but also benefit individuals in their personal and professional lives.

LOL. I don't know if this was posted ironically or not.

Writer.com: 68% human-generated content
gptzero.me: "Your text is likely to be written entirely by AI"
contentscale.ai: 40% human - "Likely to be AI generated!"
writefull.com: "44% likely this comes from GPT-3, GPT-4 or ChatGPT."
paraphrasingtool.ai: "It's likely that AI wrote this text"
hivemoderation: 99% - "The input is: likely to contain AI Generated Text"

This Hive Moderation (https://hivemoderation.com/ai-generated-content-detection) one seems to have never been wrong thus far.

The first paragraph is changed a bit by the poster. Looks like ChatGPT said "While AI tools can be helpful..." because of the comma after "errors". Otherwise the sentence and comma placement doesn't make complete sense.

The second paragraph seems to be 100% AI-written.

Look at all of this person's recent posts -- they seem plagiarized or at least highly-"assisted". All their non-bounty posts are all ploys for merit and totally unnecessary.

I thought an user who use AI generated text will avoid to reply on Meta especially this thread when it discuss about the particular matter, but this user prove me I was wrong :D

Maybe he created this thread Clarification on Plagiarism Rule and Policy (https://bitcointalk.org/index.php?topic=5449355.msg62113449#msg62113449) in order to see if AI generated post is considered as plagiarism in this forum or not.

Although there are many users have a good English skill or they're a native speaker, but I don't see they're encouraging or correcting other users grammars, moreover they can post in their own local language board. Which is off topic if he say this community help other people to improve their language skills.

Lol😆... It's funny how everybody on this forum is now an A.I suspect. Hahaha, because who knows if someday I might not be allegedly accused of being an A.I. But one thing we all need to understand is that this A.I tools we all use today weren't developed by aliens, but by humans like us, which means that there may be many persons out there that can effortlessly write some of the post we classified as 100% A.I generated contents here on this forum.

So on that note, I wish to beg whichever admin that might be placed in charge of executing judgement to whoever reported as an A.I poster to please endeavour to do thorough research. Thanks

I doubt it. Can you find me one example of a human-written post that has returned "100% AI" or even "likely AI" in 4 or more of these detectors?


Most of the critics of AI detectors use hyperbole to make their case, like the OP of this thread for example. Sure, the AI detectors are not perfect, and some are better than others.

Here's the thing: just be human and don't plagiarize or post spam, and you won't have anything to worry about. Doesn't seem like too much to ask.

Yep.  And when it's the other way around (good English suddenly turning into poorly-written gibberish) it's also a bright red flag to members who've been here a while and are well aware that bitcointalk accounts can be quite valuable and that they change hands more often than they should.

Honestly, from my experience it's extremely rare for new members to be really proficient in English, and I think that's been true since I registered in 2015.  And we all know why: sig campaigns and bounties, which pay for posts in English, draw members who need to earn money in droves--and I'd venture to guess that most are from Asian or African countries where English isn't most people's first language. 

Note: I'm not knocking anyone for wanting to participate in campaigns or even for writing in a language that's not their native one.  The problem I've always had is that poor writing is often combined with a glaring lack of thought or effort.  You'd think that if you want to maximize your earnings long-term you'd do your best when posting, but Jesus....just look at some of the posts in Altcoin Discussion or other spam magnet sections.  So many of the lousy English writers aren't even trying.

Umm... taking a risk of sounding nosy... am I wrong if I wondered what the actual purpose of this thread is? OP seems have no interest to engage on the discussion occurred by raising this thread, as I can see him posting anywhere else but here. I initially thought he's concerned about having to watering down his english and cases of "false positive" on written-by-AI detection, but that's most likely not the case as he doesn't even bother to engage to any replies here.

This has already becoming a debatable topic and I am also of the opinion that if you are using AI generated tools like ChatGPT to copy content and directly pasting it as your post it's plagiarism because you have contributed nothing in it so how come someone can believe he it's right to do so?

The AI tools also gathers information from the different resources and present you a text for the keywords you have entered as per your search but they also come from different authors so in reality you are copying someone work so it's violation of the rules.Some members do this just to earn merits but in reality they will be caught because we will have more such plagiarism detecting AI tools and moreover we can indentify to some extent the difference between AI generated posts and human written one's so there's no fun in doing so.

The writer is not the primary concern here, the content is. It makes no difference if the owner plagiarizes content or whoever works for them does.

If I have a site and you write for me, I don't need to tell the world that I employed you to write the text. I can post it in my own name or have an editor do it. Copywriters even sign contracts stating the content belongs to the site. If the text you/me/the editor wrote is plagiarized, it becomes a problem. And again, the world is not going to care if I wrote it, I asked you to do it, or one of my neighbors did.   

AI bots don't come up with their own content. They have a library of information they can refer to and pick what you asked them. A bot is a search engine writing content that has already been written somewhere before. It can't do things that have not already been done. Or in this case written. It can come up with variations, but based on some originals it has in its memory.

You don't have to post an actual link to an article you copied/paraphrased from. Naming the service is the next best thing. Source: Google, Chat GPT, Twitter, YouTube Live. No one can accuse of plagiarizing if you do that. Always post a link if you can, but since we are talking about bots here, it's not possible.   

Main problem I see is not in that some users can use AI for posting, but that we are facing up to a new low quality posts source which is not so easy to catch. Those ones who are suspected in AI posting usually leave low sense posts. It can be done by someone who understands not much and prone to verbiage, but can be AI written as well. So, as I see the situation, we are trying to find a tool to separate those ones who uses AI for making low quality content from people who can learn and improve their posts with time. Because otherwise forum will return to pre-merit situation, when it was just impossible to carry on a conversation because of one line posts written only to comply with the rules of bounty of 100-150 characters, but now it could be AI posts.

Forum is for talking to each other, for sharing experience and info, for asking questions and answering them. And everything preventing that like bad AI posts kills forums.