Bitcointalk post checker for AI

[Comeacross]

One of the challenges that newbies normally face in this forum is the use of AI to make posts. The forum is against this practice and you risk getting tagged or being ignored by many users. Some posts are not really Ai copy and paste but they involved repetitive or common Ai phrases and it's assumed to be Ai copy paste. This tool flagged those words for you so you can replace them with something else specific.

The purpose of this tool is to increase user's post effort and minimise the use of AI to make generic posts. This tool will help you to identify regular/common AI phrases in your post before posting. It suggests you replace the flagged phrases with your own specifics. It won't rewrite your post for you.

For now, I only include about 100 common AI phrases. I may update it later based on feedback from the community. There is score from 0-100 and rating between low, average and good. You lose points per flagged phrase in your post. This penalty varies by board. You lose 15 points per flagged phrase in technical discussion board and 12 points in other boards.

The tool also check for post length. You lose 20 points for writing lesser than 30 words. This is to discourage one line posts with little or no substances.

It also allows you to copy your post for AI check directly by clicking on copy for AI check button. Note: there's no built in AI checker, it only copy your post to your keyboard so you can paste it on any AI checker platform as suggest on the prompt message.

How to use the tool:
You need to download tampermonkey extention in your browser from Chrome webstore. Link to download tampermonkey extention
After installing, click on create a new script
Copy the following code and insert it.

Code:

// ==UserScript==
// @name Bitcointalk Post Checker v1.4
// @namespace bitcointalk.org
// @version 1.4
// @description Check posts for AI common phrases before posting on Bitcointalk
// @match https://bitcointalk.org/index.php?action=post*
// @match https://bitcointalk.org/index.php?action=reply*
// @grant none
// ==/UserScript==

(function() {
    'use strict';

    const AI_PHRASES = [
        "it's important to note", "let's dive in", "in conclusion", "navigating the world of",
        "in today's", "delve into", "unlock the", "embark on", "realm of", "it's crucial to",
        "as an ai", "i am an ai", "it's worth noting", "in the realm of", "at the end of the day",
        "in a nutshell", "cutting-edge", "game-changer", "ever-evolving", "tap into",
        "harness the power", "landscape of", "fast-paced world", "digital age", "paradigm shift",
        "seamless", "robust", "leverage", "optimize", "enhance your", "foster a", "it's imperative",
        "underscores the importance", "pivotal role", "testament to", "revolutionize", "unlocking",
        "it's undeniable", "needless to say", "it goes without saying", "in order to", "utilize",
        "facilitate", "streamline", "synergy", "dynamic", "holistic", "innovative solution",
        "state-of-the-art", "next-generation", "user-centric", "scalable", "empower", "transform",
        "disruptive", "unparalleled", "groundbreaking", "revolutionize the way", "key takeaway",
        "deep dive", "at its core", "in essence", "without a doubt", "it is evident that",
        "in the ever-evolving", "rapidly evolving", "constantly changing", "ever-changing",
        "game-changing", "breakthrough", "next level", "take it to the next level", "push the boundaries",
        "unlock new possibilities", "open up new opportunities", "drive innovation", "stay ahead",
        "stay ahead of the curve", "future-proof", "future of", "world of", "realm of possibilities",
        "power of", "potential of", "importance of", "significance of", "impact of",
        "role of", "journey of", "path to", "road to", "key to", "secret to",
        "ultimate guide", "comprehensive guide", "complete guide", "everything you need to know",
        "what you need to know", "here's what", "let's explore", "let's take a look",
        "let's break it down", "let's discuss", "let's talk about", "it's time to",
        "now more than ever", "more than ever before", "in today's world", "in today's landscape",
        "as we move forward", "looking ahead", "moving forward", "going forward",
        "in this article", "in this post", "in this guide", "in this piece"
    ];

    const BOARD_RULES = {
        'Technical Discussion': { penalty: 15, goodThreshold: 85 },
        'Speculation': { penalty: 12, goodThreshold: 80 },
        'default': { penalty: 12, goodThreshold: 75 }
    };

    function getBoardName() {
        const breadcrumb = document.querySelector('.navigate_section.breadcrumbs');
        if (!breadcrumb) return 'default';
        const text = breadcrumb.innerText;
        for (let board in BOARD_RULES) {
            if (text.includes(board)) return board;
        }
        return 'default';
    }

    function injectChecker(textarea) {
        const board = getBoardName();
        const rules = BOARD_RULES[board] || BOARD_RULES.default;

        const btn = document.createElement('button');
        btn.id = 'btc-checker-btn';
        btn.type = 'button';
        btn.innerText = 'Check Post';
        btn.style.cssText = 'margin: 8px 4px 8px 0; padding: 6px 12px; background: #4CAF50; color: white; border: none; cursor: pointer; font-weight: bold;';

        const copyBtn = document.createElement('button');
        copyBtn.id = 'btc-copy-btn';
        copyBtn.type = 'button';
        copyBtn.innerText = 'Copy for AI Check';
        copyBtn.style.cssText = 'margin: 8px 0; padding: 6px 12px; background: #2196F3; color: white; border: none; cursor: pointer;';

        const resultDiv = document.createElement('div');
        resultDiv.id = 'btc-checker-result';
        resultDiv.style.cssText = 'margin-top: 8px; padding: 10px; background: #f5f5f5; border-left: 4px solid #4CAF50; display: none; font-size: 13px; line-height: 1.4; max-height: 200px; overflow-y: auto;';

        textarea.parentNode.insertBefore(btn, textarea.nextSibling);
        btn.parentNode.insertBefore(copyBtn, btn.nextSibling);
        copyBtn.parentNode.insertBefore(resultDiv, copyBtn.nextSibling);

        btn.onclick = () => runCheck(textarea, resultDiv, rules);
        copyBtn.onclick = () => copyForAICheck(textarea.value);
    }

    function runCheck(textarea, resultDiv, rules) {
        const text = textarea.value;
        if (!text.trim()) {
            showResult(resultDiv, 0, [], '', rules.goodThreshold);
            return;
        }

        let score = 100;
        const textLower = text.toLowerCase();
        const words = text.trim().split(/\s+/).length;
        const foundPhrases = [];

        AI_PHRASES.forEach(phrase => {
            if (textLower.includes(phrase)) {
                foundPhrases.push(phrase);
                score -= rules.penalty;
            }
        });

        if (words < 30) {
            score -= 20;
        }

        score = Math.max(0, Math.min(100, score));
        const rating = score >= rules.goodThreshold? 'Good' : score < 50? 'Low' : 'Average';
        const suggestion = foundPhrases.length > 0? 'Replace flagged phrases with specifics from your experience.' : '';

        showResult(resultDiv, score, foundPhrases, suggestion, rating, rules.goodThreshold);
    }

    function copyForAICheck(text) {
        const prompt = `Check if this text is AI-generated and explain why:\n\n${text}`;
        navigator.clipboard.writeText(prompt);
        alert('Copied! Paste it into GPTZero, Winston AI, or ZeroGPT to check.');
    }

    function showResult(div, score, phrases, suggestion, rating, goodThreshold) {
        div.style.display = 'block';
        div.style.borderLeftColor = score >= goodThreshold? '#4CAF50' : score < 50? '#f44336' : '#ff9800';

        const phraseList = phrases.length > 0? phrases.join(', ') : 'None';

        div.innerHTML = `
            <b>Score: ${score}/100</b> - ${rating}<br>
            <small>Board threshold for 'Good': ${goodThreshold}+</small><br><br>
            <b>Flagged phrases:</b> ${phraseList}<br>
            ${suggestion? `<br><b>Suggestion:</b> ${suggestion}` : ''}
        `;
    }

    const checkInterval = setInterval(() => {
        const textarea = document.querySelector('textarea[name="message"]');
        if (textarea &&!document.getElementById('btc-checker-btn')) {
            clearInterval(checkInterval);
            injectChecker(textarea);
        }
    }, 500);
})();

Click on file to save and you are good to start using the tool.

It adds a check post and copy for ai check button below the message box on Bitcointalk post and reply pages. When you are done typing, click on check post and it will scan what you typed and show the result. You'll see your score 0-100, rating (low, average and good) and the flagged phrases if any.

Sample is seen in this image.


It won't stop you from posting even if the rating is low by the way.

Criticism, observation and suggestions are welcome.

[SquirrelJulietGarden]

One of the challenges that newbies normally face in this forum is the use of AI to make posts. The forum is against this practice and you risk getting tagged or being ignored by many users.

They can read the forum's AI guidelines.

Some posts are not really Ai copy and paste but they involved repetitive or common Ai phrases and it's assumed to be Ai copy paste. This tool flagged those words for you so you can replace them with something else specific.

Honestly, I don't have such fear because if I write posts by myself, how do I have to fear of my posts will be flagged as AI-generated?

From thinking, flow of ideas, it's mine, and words chosen, sentence structure, writings are all mine too. No ways my human-written posts will be flagged as AI-generated.

Such fear is only from people who use AI for generating their posts, then try to spin words around, humanize their AI-generated posts only.
Text spinning/disguised plagiarism.

[TokenTikas]

It also allows you to copy your post for AI check directly by clicking on copy for AI check button. Note: there's no built in AI checker, it only copy your post to your keyboard so you can paste it on any AI checker platform as suggest on the prompt message.

This (AI) checker can still be useful for many people here but those who normally write and reply in their own natural way may not really need to check for (AI), so not everybody will find it necessary to use. I tested it casually and honestly, it didn’t feel that important to me overall. Even so, if someone wants to work on it properly, it’s still possible to make the (AI) detector much better and more refined.

Honestly, I don't have such fear because if I write posts by myself, how do I have to fear of my posts will be flagged as AI-generated?

That’s exactly how it should be. When you write posts using your own thinking and personal understanding, there’s naturally nothing to fear. The fear usually comes to those who depend heavily on (AI) to write their posts. Normally, people who rely too much on (AI) for posting only end up harming the reputation of the forum in one way or another. Writing a good post is not even that complicated, once you share your opinion naturally in your own style, that is already enough. There’s no real need to depend on (AI) for every single thing.

At the same time, when someone writes with their own intelligence and confidence, they can comfortably speak without worrying too much about (AI) checkers, but people who mostly post (AI)-generated content can hardly speak with that same confidence.

[Upgrade00]

This is not a challenge that newbies or older members face, it is a tool they use to cut corners. Except for language barrier, there is no other reason you should be allowing a robot handle your communication with actual real people

You also cannot mistakenly get flagged for AI content, especially when you're being eye tested, same way you cannot be mistakenly flagged for plagiarism. Those phrases will seldomly be used by users posting naturally and when they are used, it will be done so naturally, so would not look generated to anyone reading it. I personally don't like using AI checkers to confirm if a content was AI generated.

[Betwrong]

~
Criticism, observation and suggestions are welcome.

I think it can be of great help to many and you did a terrific job creating this tool, I just wanted, since you asked, to give a suggestion. In my opinion, the following phrases should be deleted from  AI_PHRASES: "in today's", "realm of", "it's crucial to", "enhance your", "it's worth noting", "ever-evolving", "needless to say", "in order to", "transform", "future of", "significance of", "impact of", "role of", "path to", "road to", "key to", "everything you need to know", "what you need to know", "here's what", "it's time to", "more than ever before". I think these phrases are hard to replace, and I think there’s no need to replace them, in the first place.

[Comeacross]

This is not a challenge that newbies or older members face, it is a tool they use to cut corners. Except for language barrier, there is no other reason you should be allowing a robot handle your communication with actual real people

Maybe you didn't understand it yet. If you're posting just for posting sake to increase count or you let a bot handle it end to end, that is cutting corners but luckily, that's not the purpose of this tool.

The script doesn't write or post anything for you. It only scan through your text and flags phrases that are very common by AI output. You do the corrections manually so the actual rewrite is still on you. That's definitely not a bot handling your communication.

But if use it to polish an AI generated post and repost, that's actually cutting corners and the tool doesn't stop that unfortunately. The good side of it is that it helps someone who wrote own post but unknowingly slipped into AI tone.

You also cannot mistakenly get flagged for AI content, especially when you're being eye tested, same way you cannot be mistakenly flagged for plagiarism. Those phrases will seldomly be used by users posting naturally and when they are used, it will be done so naturally, so would not look generated to anyone reading it. I personally don't like using AI checkers to confirm if a content was AI generated.

You may be correct but it won't mistakenly flagged you in the way you mean. The worst it can do is to flagged phrases but your score will still be good if it only flags few phrases (1-2) in your post. You'll only get low rating when your post is full of those phrases.

We share the same opinion because I don't like relying on AI checkers too because they're not 100% reliable. It's the reason why the tool doesn't claim to detect AI but only gives option to copy for AI check.

If one's goal is to post AI generated content with minimal effort, this tool won't save them but if your goal is to write your own post and you want to avoid the writing habits of AI, it's here to help you. If you test the tool and it flags nothing, it's probably not made for you but if it flags multiple phrases from your post while you didn't say anything concrete, then it's doing it job.

I think it can be of great help to many and you did a terrific job creating this tool, I just wanted, since you asked, to give a suggestion. In my opinion, the following phrases should be deleted from  AI_PHRASES: "in today's", "realm of", "it's crucial to", "enhance your", "it's worth noting", "ever-evolving", "needless to say", "in order to", "transform", "future of", "significance of", "impact of", "role of", "path to", "road to", "key to", "everything you need to know", "what you need to know", "here's what", "it's time to", "more than ever before". I think these phrases are hard to replace, and I think there’s no need to replace them, in the first place.

Lol I got the sarcasm but thank you 

Before building the tool, I pulled over 200 posts that got called out as AI generated on various boards. Those phrases contained in the tool show up constantly in most of those posts. Having one or two in your post is harmless but having several combinations of them is probably a pattern.

Every phrase in the tool is valid in technical context but they are mostly overused by AI. I appreciate your suggestion but I will maintain the list for now. However, I might reduce the penalty to 5 points per match phrases instead of the earlier 15 and 12 points. It's important to note that the score and suggestion are just advice and not a rule. So the final call is still yours and the forum community at large.

[LoyceV]

This tool will help you to identify regular/common AI phrases in your post before posting.

This is the dumbest thing I've read today. What you call "AI", is actually a language model. That language model plagiarizes and clones human text to look as human as possible. And now you're telling me I should adjust the text I type so it doesn't look like the copycat that's trying to clone my texts? Or did you create this to "humanize" chatbot verbal diarrhea?

Your font is drunk.

[joker_josue]

It has been more than proven that these systems fail more often than they succeed.
AI systems (hence the name) are constantly improving their writing style and enhancing their naturalness to appear as human as possible. Today, it's almost faster to say that a human text was created by AI than to say that a text was actually created by AI.

If someone gets caught using AI text, it's clearly because they are misusing this tool.

They want to know that you use AI, they interact with the person, and things change completely.

This reminds me of something that happened to me 20 years ago, while I was still studying.

A teacher, in a tone of positive criticism, told me to look at another student who got a higher passing grade on a test than I did (I had already gotten a good grade).
I approached the teacher and asked if that student knew the answers to the test questions that day. Because knowing something is one thing, but memorizing it for a specific moment is another.
The professor chuckled approvingly and simply told me to at least try to be less "joking around" in class.

In that same class, the professor raised a problem, and no one was answering it. Me and another colleague were just chatting with each other (we were kind of mischievous).
Despite this, and after some time without responding, I stop the conversation with my colleague, answer the problem, and move on. The answer was correct.
Finally, I commented to the professor that, after all, the classmate with the best grade didn't seem to know the material. He laughed again and told me "not to be so mean."

Finally, I admit that I was never a great student. But I told this story to explain that it's through interaction that you see who a person really is.

Unfortunately, many insist on evaluating others by metrics rather than by interaction.

[bitbollo]

The real problem isn't the use or not of AI. The real problem is... the final outcome.
If you get a boring wall of text full of errors, and text that pop up from nowhere... well that's real poor content.
If you clearly can't manage an argument and rely enterely on the machine you will never be at the same level of the ones that real know it.
Another tool for hiding shitpost? I could say, no thank you. But at the end we can decide who to trust / ignore and so on...

[Comeacross]

This is the dumbest thing I've read today. What you call "AI", is actually a language model. That language model plagiarizes and clones human text to look as human as possible. And now you're telling me I should adjust the text I type so it doesn't look like the copycat that's trying to clone my texts? Or did you create this to "humanize" chatbot verbal diarrhea?

If the tool is dumb, I'll own that. It's just a simple keyword filter intend to help people who want to learn how to post here without getting blamed for sounding like a bot. I'm not claiming this is a solution by the way.

It sounds bad to me when you put it that I want you to adjust your text to not look like a copy cat but I agree. My intention was not to make you write like a bot pretending not to be a bot, my intention was to avoid a style that get your posts suspicious for being a generated content regardless of who wrote it. It's not about hiding AI use.

Sorry, but I wasn't trying to humanize chatbot. The tool doesn't write anything on your behalf. It does not call any API or turn any AI text into something that looks human. It only scan your post and flag phrases that show up constantly in the low effort posts usually made by AI.

I really appreciate the push back because it's the only way to get better. I really don't want the tool to become what's it's meant to prevent.

[SquirrelJulietGarden]

And now you're telling me I should adjust the text I type so it doesn't look like the copycat that's trying to clone my texts? Or did you create this to "humanize" chatbot verbal diarrhea?

I thought about the same utility of this tool.

Months after AI explosion and massively used for plagiarism, there was appearances of Humanized AI tools, and this one serves similar use case. Because if a poster writes posts by himself, there is no fear of getting flagged and reported as AI-generated posts.

Such fear is only from people who use AI for generating their posts, then try to spin words around, humanize their AI-generated posts only.
Text spinning/disguised plagiarism.

Such as in this discussion

humanize AI-generated content

This is one of the biggest misnomers to be applied to any type of new technology. The idea that a machine attempting to emulate a human being can further "humanize" its output is absurd. The machine is doing all the work -- there is no human beings involved in "humanization." Thus a more proper term would be "shitifier" as it makes output more shitty albeit in a very robotic way.

As far as the subject of this thread is concerned, I don't think the mods want more work of having to go out of their way to determine what is or isn't AI content. It is already hard enough to get some posters to admit they are using AI. The ultimate offense that determines if a post should be removed is whether or not it falls into the category of "spam" in the eyes of the moderators.

[LoyceV]

my intention was to avoid a style that get your posts suspicious for being a generated content regardless of who wrote it.

It's the same thing, if anyone wants to be suspicious: let them. There's a very simple way to not be suspected of using spambots to write posts: just don't do it.
If a chatbot copies my style, there's nothing I can do about that. Changing my style is utterly pointless as the chatbot can (and will) just copy it again.

It only scan your post and flag phrases that show up constantly in the low effort posts usually made by AI.

So self-censorship because of chatbot existence. No thanks.

[Alpha Marine]

It's just a simple keyword filter intend to help people who want to learn how to post here without getting blamed for sounding like a bot. I'm not claiming this is a solution by the way.

I have never once stopped to think if my post sounds AI-generated. I have never used AI to generate a post on this forum, not even a line of a post, so why would I bother? If I use certain words or phrases that look suspicious to another person, then that is not my problem.
I can't be looking out for for keywords not to use simply because I don't want to sound like a bot.

I feel like people make the forum too formal. This is not a board meeting or a law office. You don't need to write like you're trying to impress your professor. Just write the normal way you would write in an informal text. For me, as long as your English is okay, no need for big grammar and phrases.

For people that English is not their first lanaguage and claim to use AI to make their English better, there are tools that corrects spelling and bad grammer, so I really don't buy that excuse of usng AI to generate post simply because their English is not good.

[mindrust]

This script is a solution that looks for a problem which doesn't really exist.

I can talk like AI since I have been spending so much time with it lately but no matter how much I try, I can't fully replicate AI responses and trigger forum's AI detectors. I don't think anybody can. Even if English is your native language, you still cannot do that imo.

That doesn't mean you should stop developing. Just, this is not it. Create something else.

[rat03gopoh]

I really appreciate the push back because it's the only way to get better. I really don't want the tool to become what's it's meant to prevent.

Since this tool clearly has the potential for abuse, it's better not to launch it in the first place instead of improving it, sorry.
In case you didn't know, the admin handles reports related to AI posts manually, even though there are many AI text detectors out there.