@Lucius accusations: is there hope for all newbies?

[dkbit98]

I am really embarrassed at what I saw. I am willing to share my hive account username to anybody except L...

Oh look I found another post that is 100% FAKE according to exact same stupid AI Sapling tool:

I think that Ledger has already confirmed that part or even the whole device actually consists of parts that are manufactured in China, so it is completely meaningless whether something was manufactured in China or the parts were imported from China and assembled somewhere else. I have some Chinese devices that are of premium quality and have been working without any errors for years, so I have no doubt that the Chinese cannot make a high-quality hardware wallet, the only question is whether it is completely open source and whether they might have a backdoor.

Not that I want to scare anyone now, but it is a well-known fact that the Chinese have been caught several times in the act of military and industrial espionage by installing hidden chips in computer components. This is why the US administration started a trade war with them and decided that all key components for computers/servers used in critical infrastructure will be produced outside of China.

https://bitcointalk.org/index.php?topic=5479306.msg63403204#msg63403204



I am not going to blindly trust AI detection tools like this because I know they are stupid, and they can have false positives for everyone including me.
Not that I am defending or attacking anyone, just saying what I observed when I used tools like this.
Best results can be achieved when you using multiple tools to check if someone is using AI or not.

[1714693166]

1714693166

[1714693166]

1714693166

[1714693166]

1714693166

[1714693166]

1714693166

[1714693166]

1714693166

[1714693166]

1714693166

[Lucius]

I have no choice but to apologize to the OP, because I checked his post with only one tool that I was sure could successfully detect content generated with AI, given that many people use it in the thread where AI generated posts are published. Obviously, it is a tool that no longer makes sense to use because it detects most posts as if they were created with AI.

[BitcoinGirl.Club]

I have no choice but to apologize to the OP, because I checked his post with only one tool that I was sure could successfully detect content generated with AI, given that many people use it in the thread where AI generated posts are published. Obviously, it is a tool that no longer makes sense to use because it detects most posts as if they were created with AI.

I don't blame you. Looking at how people using AI to make generic response, anyone can suspect a long easy and check with the tools. Fact is this tools are not 100%. They use algorithm which are not perfect. If you check same text with different tools, you will find different result. On the other hand AI is becoming smarter which is also creating confusions for the tool developers.

[examplens]

I have no choice but to apologize to the OP, because I checked his post with only one tool that I was sure could successfully detect content generated with AI, given that many people use it in the thread where AI generated posts are published. Obviously, it is a tool that no longer makes sense to use because it detects most posts as if they were created with AI.

I don't blame you. Looking at how people using AI to make generic response, anyone can suspect a long easy and check with the tools. Fact is this tools are not 100%. They use algorithm which are not perfect. If you check same text with different tools, you will find different result. On the other hand AI is becoming smarter which is also creating confusions for the tool developers.

A skilled copywriter will easily "tweak" the text so that it looks more like it was written by a human. Honestly, that's what it looks like the most to me.
That the OP is being helped by AI tools, dkbit98 found in his other post, which only strengthens my suspicions.

[nutildah]

I have no choice but to apologize to the OP, because I checked his post with only one tool that I was sure could successfully detect content generated with AI, given that many people use it in the thread where AI generated posts are published. Obviously, it is a tool that no longer makes sense to use because it detects most posts as if they were created with AI.

It still makes sense, it just generates false positives from time to time. That's why in the thread I have this clause:

At least 2 of the 6 AI content detectors mentioned below must give a result of the post having a strong likelihood of containing AI-written material (with "strong likelihood" defined for each one):

1. Hive Moderation* - 70% likely to contain AI generated text or above
2. Writefull GPT Detector - 70% likely to come from GPT-3, GPT-4 or ChatGPT or above
3. Writer AI Detector - 30% human-generated content or below
4. Copyleaks AI Content Detector - 70% probability for AI or higher
5. Contentatscale AI Detector - 30% likely to be human or below
6. Sapling AI Detector - 70% Fake or above

I've found the 3 best ones to be Hivemoderation, Copyleaks & Sapling.ai... but you can't rely on just one. Ideally 3 should be used if there's any doubt whatsoever. There's also this other one I forgot to try out but apparently it does a decent job as well:

To achieve 100% accuracy check the content using Easy AI Cheker https://easyaichecker.com/

A skilled copywriter will easily "tweak" the text so that it looks more like it was written by a human. Honestly, that's what it looks like the most to me.

This also could be the case. But if the author is going through the trouble of doing some re-writing, that's OK with me personally, as it means they at least put some thought & time into it. If just one or two words are changed, they add the words "I personally believe" at the beginning, or they add extra punctuation or something cheesy, that's still bad.

[BitcoinGirl.Club]

A skilled copywriter will easily "tweak" the text so that it looks more like it was written by a human. Honestly, that's what it looks like the most to me.

This also could be the case. But if the author is going through the trouble of doing some re-writing, that's OK with me personally, as it means they at least put some thought & time into it. If just one or two words are changed, they add the words "I personally believe" at the beginning, or they add extra punctuation or something cheesy, that's still bad.

That's Symmetrick Ratimov was doing but for the OP, we will never know.

[FatFork]

A skilled copywriter will easily "tweak" the text so that it looks more like it was written by a human. Honestly, that's what it looks like the most to me.

This also could be the case. But if the author is going through the trouble of doing some re-writing, that's OK with me personally, as it means they at least put some thought & time into it. If just one or two words are changed, they add the words "I personally believe" at the beginning, or they add extra punctuation or something cheesy, that's still bad.

This technology's just going to keep affecting things more and more.  The quality of content on forums and such is gonna change big time as these generative AIs evolve and get better, and they will.  AI detection tools catching bot writing already proven unreliable by giving false positive results.  Plus there's already tools rewriting bot posts so they'll trick AI detection.  Not sure how we actually stop this down the road.

[Lucius]

I don't blame you. Looking at how people using AI to make generic response, anyone can suspect a long easy and check with the tools. Fact is this tools are not 100%. They use algorithm which are not perfect. If you check same text with different tools, you will find different result. On the other hand AI is becoming smarter which is also creating confusions for the tool developers.

It is said that whoever works makes mistakes, and in the future I will have to be much more careful when it comes to AI chat bots and public accusations for those who may be using them. I know I first used the disputed detector and it showed me over 70% AI content, then I checked with another one I found online (not entirely sure of the name) which also showed over 60%. I think you're right, AI will become so sophisticated that we will hardly be able to recognize whether people or bots are writing here.

A skilled copywriter will easily "tweak" the text so that it looks more like it was written by a human. Honestly, that's what it looks like the most to me. That the OP is being helped by AI tools, dkbit98 found in his other post, which only strengthens my suspicions.

What exactly do you mean? I just see that he quoted one of my posts, which the tool also detected as an AI generated post. Of course, it is possible to additionally disguise such posts in such a way that they are a combination of AI, paraphrasing and plagiarism.

~snip~
This also could be the case. But if the author is going through the trouble of doing some re-writing, that's OK with me personally, as it means they at least put some thought & time into it. If just one or two words are changed, they add the words "I personally believe" at the beginning, or they add extra punctuation or something cheesy, that's still bad.

There are definitely those who completely plagiarize or use AI, and I don't think that any justifications should be sought for them, but is it even possible to determine what is the permissible level of using AI tools, without having to react to it? A Legendary member was caught by you not so long ago using AI, although not in the worst way - but still in a way that his posts were not entirely original, right?