My AI experiment on the forum

[dkbit98]

From now, seoincorporation isn't a part of the campaign until he can be able to remove the neutral tag from his profile.

Neutral tag should not be a problem for joining signature campaigns, it's just showing what happened in the past with that member.
I doubt anyone will remove their neutral tags for him, but this is not the same as having negative tag and he was not accused for scamming anyone.

[Zwei]

but i want to make noise to make us all understand where we are with AI, and understand that this is a fight that we can't win.

except the AI lost in this case.

i want to believe what you are saying, but at the same time, it's hard to believe.
if you really just wanted to understand where we are with AI, you could have just started a discussion thread like a normal person would do and showed there that you can create posts in your style or whatever your experiment was about.
it would have been much better that way instead of this clusterfuck you created for yourself.

like how in the hell did you think this was a good idea? how did you think it's gonna end for you? like we would just go, cool AI experiment bro, you totally did not do it just to get paid... so don't do it again.
for someone who been here since 2014, i expect you to know better than this.

and let's not forget the fact that you did it while being in a sig campaign, and you didn't even tell LM about it in advance.

I'm ready for the consequences, if you want to ban me or burn my reputation for this experiment, do it. I was totally conscious of what i was doing and enjoyed the experiment.

i don't know if you really mean that, or if it's just some reverse psychology.

[Alone055]

Neutral tag should not be a problem for joining signature campaigns, it's just showing what happened in the past with that member.
I doubt anyone will remove their neutral tags for him, but this is not the same as having negative tag and he was not accused for scamming anyone.

Even though you're right, I would say that when a user gets a neutral tag that says a user is either simply a spammer or an AI-spammer during the time they are in a paid signature campaign, it basically says that they have been misusing the opportunity by spamming around the forum only to make money from the campaign and haven't been producing anything much organic. So basically, it's not only the scammers that shouldn't get slots in signature campaigns.

I know OP has been a very good user in the past, and he isn't generally a spammer, but he did make a mistake by using an AI model to make posts while being in a signature campaign, and I guess this is his punishment for doing that. He wouldn't have any problem, and this discussion or thread would have had a different perspective if OP had created it before doing the so-called experiment, as he calls it, instead of creating it after he was caught and reported by users for making AI-generated posts.

[Stalker22]

~
He wouldn't have any problem, and this discussion or thread would have had a different perspective if OP had created it before doing the so-called experiment, as he calls it, instead of creating it after he was caught and reported by users for making AI-generated posts.

Exactly my opinion as well.  Besides, he really should ave talked things through with his manager first before trying that "AI experiment" since Im pretty sure most managers dont want AI used for pay-to-post campaigns. But he just went ahead without asking.  Not smart.

[seoincorporation]

That's unfortunate to see seoincorporation did this.

I have a habit of inviting users in DM to join my campaigns. I find users based on my needs, targeted boards etc. seoincorporation is one of them.

I can confirm this, LM send me the proposal to join the campaign, and the users on his campaigns are carefully selected users.

What surprised me the most is that he barely posted more than 20 posts per week. That's why when he sent me a DM regarding this issue, I trusted his words. It makes sense if you ask me. If he were a sig spammer, why would he post fewer than 20 posts per week despite having the option to get paid for 35 posts? BTW, he didn't tell me this. This is my observation because I remember I always wanted him to post more, but have always been disappointed by his numbers lol.

From now, seoincorporation isn't a part of the campaign until he can be able to remove the neutral tag from his profile.

And here you have a good point, i really wish to be able to post more in your campaign but my full time job really absorve my time, working from 9am to 7pm from Monday to Saturday is complex, even those 20 post a month that i used to do wasn't easy but i did my best to stay active on the forum. Even with AI i wasn't able to fill the 35 posts because it wasn't full auto i used to read the post and to verify the answer, so i was conscious about the threads. But the income from this campaigns didn't make any difference to me, i used that money to gamble most of times, so, isn't a big deal for me to lose the signature.

I know there would be consequences from my actions, and I'm not even angry with LoyceV, in fact i admire him for all the contributions he/she has made to the forum, so, no hard feelings and i respect his decision. Even if his neutral trust has a lot of hate there is not problems.

This signature spammer posted chatbot verbal diarrhea AKA plagiarism and claims "it was an experiment". Don't waste your time interacting with it.
Second link: https://asktom.cf/index.php?topic=5546497.msg65474764#msg65474764

If any of you think the same way i hardly recommend to use the ignore button that you can find on any of my posts.

I'm still thinking about ways to fight against AI in the forum and feeling kind of stuck but maybe working with the community we could make the right tool for it. My idea is the next one:

We have the patrol tool: https://asktom.cf/index.php?action=recent;patrol

So, we could make an script that parce that page and send the post to an IA agent where the post could get cataloged like "Legit, Random, Spam, Scam, IA..." And that way we could manually verify each list, that process could be full auto with a delay on the call of patrol to avoild getting IP banned.

But the problem is that patrol only show new posts, and if we want to monitorate the full forum we should filter the replies on the threads too, which is complex because the RSS is blocked and if we abuse the forum calls to verify the different posts we will get blocked by the server, so, if someone has a way to get all the new posts on all the forum that would help to build the right patrol tool.

The IA agent could include the ZeroGPT API and other ones to filter the posts, and GPT to identify scams.

But the problem is that patrol only show new posts, and if we want to monitorate the full forum we should filter the replies on the threads too, which is complex because the RSS is blocked and if we abuse the forum calls to verify the different posts we will get blocked by the server, so, if someone has a way to get all the new posts on all the forum that would help to build the right patrol tool.

As linus torvalds say... Talk is cheap, show me the code.

Code:

from bs4 import BeautifulSoup
import json
import sys
import requests
import re

def extract_after(text, key):
    try:
        return text.split(key)[1].split()[0]
    except:
        return None

def parse_quote_header(header_text):
    match = re.search(r"Quote from:\s*(.+?)\s+on\s+(.*)", header_text)
    if match:
        return match.group(1).strip(), match.group(2).strip()
    return None, None

def extract_user_profiles(soup):
    profiles = {}
    for td in soup.find_all("td", class_="poster_info"):
        a = td.find("a")
        if a:
            name = a.text.strip()
            href = a.get("href")
            profiles[name] = href
    return profiles

def extract_quotes_recursive(container, user_profiles):
    quotes = []
    headers = container.find_all("div", class_="quoteheader", recursive=False)

    for header in headers:
        quote = {}
        link_tag = header.find("a")
        quote["link"] = link_tag["href"] if link_tag else None
        user, date = parse_quote_header(header.get_text(strip=True))

        quote["author"] = user
        quote["profile_url"] = user_profiles.get(user, None)
        quote["date"] = date

        quote_block = header.find_next_sibling("div", class_="quote")
        if quote_block:
            quote["quotes"] = extract_quotes_recursive(quote_block, user_profiles)
            for q in quote_block.find_all("div", class_="quote", recursive=False):
                q.decompose()
            quote["content"] = quote_block.get_text(strip=True)
            quote_block.decompose()
        else:
            quote["quotes"] = []
            quote["content"] = ""

        header.decompose()
        quotes.append(quote)

    return quotes

def parse_html_posts(html_content):
    soup = BeautifulSoup(html_content, "html.parser")
    post_blocks = soup.find_all("td", class_="msgcl1")
    user_profiles = extract_user_profiles(soup)
    posts_data = []

    for block in post_blocks:
        post = {}
        anchor = block.find("a")
        post["message_id"] = anchor.get("name") if anchor else None

        poster_td = block.find("td", class_="poster_info")
        if poster_td:
            user_link = poster_td.find("a")
            post["author"] = user_link.text.strip() if user_link else None
            post["profile_url"] = user_link["href"] if user_link else None

            activity_text = poster_td.get_text()
            post["activity"] = extract_after(activity_text, "Activity:")
            post["merit"] = extract_after(activity_text, "Merit:")

        subject_div = block.find("div", class_="subject")
        post["title"] = subject_div.get_text(strip=True) if subject_div else None

        date_div = subject_div.find_next_sibling("div") if subject_div else None
        post["date"] = date_div.get_text(strip=True) if date_div else None

        post_div = block.find("div", class_="post")
        if post_div:
            post["quotes"] = extract_quotes_recursive(post_div, user_profiles)
            post["content"] = post_div.get_text(strip=True)

        posts_data.append(post)

    return posts_data

def main():
    if len(sys.argv) < 2:
        print("Usage: python3 post_last.py <URL> [output.json]")
        sys.exit(1)

    url = sys.argv[1]
    output_path = sys.argv[2] if len(sys.argv) > 2 else "bitcointalk_parsed.json"

    try:
        headers = {"User-Agent": "Mozilla/5.0"}
        response = requests.get(url, headers=headers)
        response.raise_for_status()

        posts_json = parse_html_posts(response.text)
        with open(output_path, "w", encoding="utf-8") as outfile:
            json.dump(posts_json, outfile, indent=2, ensure_ascii=False)

        print(f"Success! Saved to {output_path}")

    except requests.RequestException as e:
        print(f"Error fetching URL: {e}")
        sys.exit(1)

if __name__ == "__main__":
    main()

Output:

https://privatebin.net/?b23d0b444e13d295#5xcuFNDVwzcPdZBjaiJtZoqaijrxUsstVM1G98WycE8z

Run it:

Code:

python3 post_last.py https://asktom.cf/index.php?topic=5546497.0 out.json

With this code we can directly parsing any thread link from the forum and get a JSON of the posts and quotes, that would be a nice input for an automated process to filter the content. A tool like this could be a nice base to patrol the forum, with another script we could get the updated threads directly from the board of our interest, then create all the JSONs and feed the AI agent.

[Free Market Capitalist]

I'm still thinking about ways to fight against AI in the forum and feeling kind of stuck but maybe working with the community we could make the right tool for it.

Haven't you thought of ideas that can help you use AI on the forum without it being a simple copy and paste? A mix of human interaction and improvement with AI.

As I said in another thread I tried in a newly created thread that I made, pass it through the AI to improve it but the problem is that it came out in the detectors as 100% AI generated or almost, when that is a false positive since the content had been created by me. When the AI made the modifications the detectors detected it as AI generated, when in fact it was only AI modified. Between that, and I was not convinced by any of the versions I was given, in the end I published it as I had written it.

[ABCbits]

Output:

https://privatebin.net/?b23d0b444e13d295#5xcuFNDVwzcPdZBjaiJtZoqaijrxUsstVM1G98WycE8z

FYI, that link currently show this error message.

Could not get paste data: Paste does not exist, has expired or has been deleted.

[lovesmayfamilis]

There used to be campaigns paying per post, these were more fun. Made 10 posts? Here xxx satoshis. Made 20? Here is double. Asking people to make 25+ posts every week is kind of creating pressure on the posters. That clearly shows it here.

The signature of the company in which the OP participated required writing ten posts in the gambling section to receive payment. But as you can see, the OP claims that it was difficult for him to participate in the company since his time was limited.
If so, who still doubts that he wanted to use AI to participate in the forum, since he had neither the time nor interest in participating in this company and used the money received only for games? I would say that this looks a little disingenuous. Why take someone's place if everything is very unimportant to you? Especially knowing that you were invited by the manager, and he hopes to see good content from you.

And here you have a good point, i really wish to be able to post more in your campaign but my full time job really absorve my time, working from 9am to 7pm from Monday to Saturday is complex, even those 20 post a month that i used to do wasn't easy but i did my best to stay active on the forum. Even with AI i wasn't able to fill the 35 posts because it wasn't full auto i used to read the post and to verify the answer, so i was conscious about the threads. But the income from this campaigns didn't make any difference to me, i used that money to gamble most of times, so, isn't a big deal for me to lose the signature.

After all that has been said, one can ask where the OP got the time to set up the AI? This means there were certain goals to improve or speed up the process of creating the required quantity for the quota.

[Free Market Capitalist]

After all that has been said, one can ask where the OP got the time to set up the AI? This means there were certain goals to improve or speed up the process of creating the required quantity for the quota.

It probably didn't take him long:

I work for a travel company that is in a automatization and AI integration process and I'm the project leader, i have done some bot with AI for Telegram, WhatsApp, Site Chats, and phone agents.

[mindrust]

Protip #2

If you are going to use AI to accomplish something which you think it is the “smarter” way,

Also ask AI’s opinion on what you are about to do.

It often gives you the correct potential consequences of your actions. Then ask yourself: “Is the reward worth the trouble I can potentially face?” If you are in doubt ask AI’s opinion one more time and see it yourself.

That way you can stop yourself from doing stupidly dangerous or shady things.

[seoincorporation]

Output:

https://privatebin.net/?b23d0b444e13d295#5xcuFNDVwzcPdZBjaiJtZoqaijrxUsstVM1G98WycE8z

FYI, that link currently show this error message.

Could not get paste data: Paste does not exist, has expired or has been deleted.

The output is a JSON of the posts, le me share a piece of the JSON, that way you have an idea about the output:

Code:

...
{
    "message_id": "msg65475559",
    "author": "bitmover",
    "profile_url": "https://asktom.cf/index.php?action=profile;u=1554927",
    "activity": "2702",
    "merit": "6688",
    "title": "Re: My AI experiment on the forum",
    "date": "June 12, 2025, 11:30:37 AM",
    "quotes": [
      {
        "link": "https://asktom.cf/index.php?topic=5546497.msg65474764#msg65474764",
        "author": "LoyceV",
        "profile_url": "https://asktom.cf/index.php?action=profile;u=459836",
        "date": "June 12, 2025, 05:50:50 AM",
        "quotes": [
          {
            "link": "https://asktom.cf/index.php?topic=5546497.msg65474546#msg65474546",
            "author": "seoincorporation",
            "profile_url": "https://asktom.cf/index.php?action=profile;u=334783",
            "date": "June 12, 2025, 02:28:11 AM",
            "quotes": [],
            "content": "I have been using AI to generate some of my posts in the past 6 days. My intention was to experiment if the forum detects the AI and how far a user can go with AI abuse."
          }
        ],
        "content": "So you're just another spammer.Thousands of users have been doing that, and it's very annoying! It's a waste of everyone's time, and posting chatbot verbal diarrhea as your own is plagiarism by definition.That's no way to treat a community!"
      },
      {
        "link": "https://asktom.cf/index.php?topic=5546497.msg65474604#msg65474604",
        "author": "nutildah",
        "profile_url": "https://asktom.cf/index.php?action=profile;u=317618",
        "date": "June 12, 2025, 03:25:15 AM",
        "quotes": [],
        "content": "Uh... pretty obviously wecanwin it as all your posts were deleted... A giant waste of your time & everyone else's IMO."
      }
    ],
    "content": "This is a tremendous waste of time, and since the user is filling signature campaign posts quota, this is wrong with the community. Why not create a new account to do such experiment?Maybe if OP had talked to the mods or admin beforehand, it would have made the experiment a little better. As OP was detected because admin/mods had already tested such AI detection tool, this is precisely the reason why it worked. No real experiment was conducted that hasn't been done before."
  },
...

After all that has been said, one can ask where the OP got the time to set up the AI? This means there were certain goals to improve or speed up the process of creating the required quantity for the quota.

If the goal was to cover the full quota (35 post/week) i would automate the full process, but that wasn't the goal. I was trying to understand where is the forum and the community in the AI topic.

And now i will build a AI tool to patrol the forum, i have a clear idea, only need the time to code it, But it will be my apport and the biggest consequence of my experiment    

[stwenhao]

but the problem is that it came out in the detectors as 100% AI generated or almost, when that is a false positive since the content had been created by me.

It is not a bug, it is a feature. If you use AI too much, then you start behaving like bots. And then, you may be detected as a bot, or have problems with CAPTCHAs.

Also ask AI’s opinion on what you are about to do.

AI can tell you everything you want to hear. It can give you a list of advantages of breaking hash functions. It can tell you that stealing content is a good thing. If your moral compass is AI-based, then it usually won't protect you from many mistakes: https://www.youtube.com/shorts/eVaj8YIS0bc

[Floxynice]

If the goal was to cover the full quota (35 post/week) i would automate the full process, but that wasn't the goal. I was trying to understand where is the forum and the community in the AI topic.

And now i will build a AI tool to patrol the forum, i have a clear idea, only need the time to code it, But it will be my apport and the biggest consequence of my experiment    

Your goal in all of this may be genuine and with pure intentions, but the problem here lies in how you went about it with no precautionary measures, as if it was never going to be a big deal if it eventually got detected.

Maybe a little description at the bottom of your posts explaining the purpose of your supposed experiment would have saved you from trouble. Now, who would believe it was indeed an experiment?

For me, this was an experiment taken too far, a very risky one.

[Alone055]

And now i will build a AI tool to patrol the forum, i have a clear idea, only need the time to code it,

You can always use an AI model to do that if you are short on time, for experimental purposes.

[Don Pedro Dinero]

A question I was asked in the local board gave me the idea of explaining here what happens in another forum where AI is allowed.

forocoches.com is the largest Spanish-speaking forum, I have an account on it and visit it every day. AI is allowed on it and I have seen it used several times, without quoting. For example, someone opens a topic related to AI and people post what the OP says in chatGPT and copy paste the answer (without quoting or acknowledging that it is AI generated, but it is clear). It's an entertainment forum that allows soft trolling and such, so such behaviors are not uncommon.

The positive thing I see is that despite being authorized the use of AI that forum has not become, far from it, a forum of soulless bots. 99% of what is written is clearly written by humans. I say this because I have seen this fear mentioned here although the essential difference is that in that forum people do not get paid for writing.

That said, I have to say that I agree with putting some curb on AI in this forum both through moderation and us members with the use of tags following nutildah's thread rules.

Although I wanted to put that point of hope that AI doesn't have to do away with human interactions in the forums by a long shot.

[qurbanshah02]

And now i will build a AI tool to patrol the forum, i have a clear idea, only need the time to code it,

You can always use an AI model to do that if you are short on time, for experimental purposes.

It is designed for this purpose and if we see that we do not have any information, we first choose AI and get help from it to do our work. It is very good and useful. It is getting even better because it will become even better with new technologies.

[nutildah]

although the essential difference is that in that forum people do not get paid for writing.

This has everything to do with it. ChatGPT is a sig spammer's dream. It makes it so easy to say nothing, so quickly, flawlessly and effortlessly. I've seen people using AI in places like Facebook where there is nothing to be gained for most people using it. Pretty much every other post shown to me is written using AI, and I gotta say, its disturbingly robotic. But if there was no monetary reward for posting here, I don't think AI would be as big of a problem as it is.

Although I wanted to put that point of hope that AI doesn't have to do away with human interactions in the forums by a long shot.

Agreed -- there will always be actual people conversing with one another here, for so long as the forum exists. A lot of accounts that turn to using AI now I think secretly wished they were bots before, so I never desired to interact with them in the first place.

[Alone055]

It is designed for this purpose and if we see that we do not have any information, we first choose AI and get help from it to do our work. It is very good and useful. It is getting even better because it will become even better with new technologies.

There is nothing wrong with using AI in anything you are doing; what's wrong is making it generate information, and then passing that on as your own, instead of admitting you've used an AI model to generate that, and you have zero contribution to that information. Even if you do have some contributions in it, but it's partly generated using AI, you are still supposed to mention that. I'm talking about this forum only, btw, but it's unethical to do that anywhere, imo.

[The Sceptical Chymist]

What would Sarah Connor do?

Bulk the fuck up and stick Theymos's neck with a syringe filled with Draino?

I've got no problem with OP's experiment, because at the very least some of us who care about bitcointalk are discussing what a nightmare AI can turn this forum into (and yeah yeah yeah I know there have been and still are threads dealing with this).  I seriously fear for the future of this beloved hangout--not only could AI turn it into a wasteland of nonsense posts, but discussion forums seem to be headed toward dinosaur land.  A lot of shitcoins don't  even use Reddit much as a place to go for information, help, to talk with the community or whatever, and if you look at some of the coins that have their own discussion forum they're absolutely dead.

Sig campaigns are keeping bitcointalk alive, whether you like it or not.  The goddamn irony is that they're also what bitcointalk is plagued with, and combined with tools to make it easier to write a literal no-brainer post, it's what's going to burn this house to the ground.  Maybe.  Hopefully not.

Don't give me shit about being in a campaign myself.  Some members aren't here for just that, but I'd say 99.7% are.

[nutildah]

Sig campaigns are keeping bitcointalk alive, whether you like it or not.  

Honestly I think credit goes to theymos for that -- he's the one that pays the bills & keeps the lights turned on. Much like Bitcoin itself, Bitcointalk doesn't need to do shit except for continue to exist and function as intended. The real value of the forum lies in the historic nature of posts by satoshi, Hal, Vitalik and other crypto luminaries. To those who are just visiting the forum as a casual reader, its stuff like this that gives Bitcointalk eternal relevance:



So basically if sig campaigns disappeared tomorrow, the forum would be fine & continue to exist. Not like its dependent on revenue to keep operating.