AI Generated Posts

[ThomasB]

Some have predicted that 90% of online posts will be AI generated within a year. And I have seen some posts on this forum that are probably created by AI.

Although it is mostly gut feeling, one general clue is a brand new member that replies to a discussion with all the right key words and is well spoken. But is vague at the same time.

@all, What is your opinion of interacting (on this forum) with AI that is impersonating as a human that needs help or is giving advice?

For the record, I don't believe that I am an AI machine. But sometimes I'm wrong, which is worrisome.

- Thomas

[Ath]

Confirming your suggestion about AI probably being used to write posts. There's a 'certain user' that I've been suspecting of using AI to generate answers and copy/pasting them in the forum, especially if a question has been unanswered for a couple of days. In some way the ignorance and often half baked answers suggest that an AI is involved, as any right-thinking person would probably understand that the content of the reply isn't usable as an answer, and not post it.

The enthusiastic reply to your other post is still under observation for possible spam-injection, but as the rather short time before editing is locked, often passes quicker than these users expect, they seem to be somewhat disappointed, and sometimes just don't return. Which is fine

[ThomasB]

A major issue is that spam-injection, whether helpful or not, is quickly becoming fully AI driven. No longer human involved, it's all machine driven.

I believe the AI devs have their AI Bots interact in forums to refine their databases. And also to establish a member profile that later will be used to promote nefarious things. Their long term goal is not in our best interests.

Which is why I am bothered when I detect them infiltrating a useful online discussion group like this one. I don't care if they are helpful, I prefer they stay off my front lawn.

In the meantime, I will continue to pinch myself. When it stops stinging I will know I have become an AI NPC. Because that's probably one of their goals.

- Thomas

[TD-er]

I know 100% for sure we have had some AI bots here on the forum.
Some of them I immediately recognize as spammers and when checking their IP/mail/username on stopforumspam.com I nearly always get confirmation.

However there are for sure spammers using botnets which do create several users where one of them asks a completely unexpected and quite often unrelated question to which another new user replies within hours/days confirming they also would like to know about that specific topic.
Often a 3rd one (also new account) then joins and quotes them using 0-size links in the reply.

Those are so extremely non-intelligent, I would rather qualify those as "spambot-intelligence" (like no intelligence at all)

Still, the spam-posts do seem to get a bit more 'on-topic' lately and I did recognize some of my own wording in a few of those posts. So it seems our forum is also used for AI-training.

Typically I do remove about 10 user-accounts for spamming per week.

You should also check their signatures as those often are filled with 0-size links.
Or when they quote a post, in the quote there will be links (0-size) or they just changed existing links to whatever they would like to spam about.

[TD-er]

The reply to your post which may have looked 'ontopic' from an AI LLM point of view has been removed.
I just checked and the spammer was now listed on stopforumspam.com

This "user" has an IP-ban, username ban and email-address ban.

Yesterday I did spend about 2 hours looking through all users which have been active in the past 2 months and found about 10 spammers in there which are now also blocked and removed.

What did struck me as being odd is that some of them had features enabled which they should not have, so maybe they also use some 0-day in phpBB (or at least not yet patched in the version we're running)
And some of them had listed an email-address with a domain which doesn't seem to have any MX-records and was not changed in quite a while.
So no idea how those got past the email-verification.

[ThomasB]

At some point it will be impossible to tell authentic human members from insidious AI members. And that time is fast approaching. But for now, fight them with all you got!

- Thomas

[ThomasB]

At some point it will be impossible to tell authentic human members from insidious AI members. And that time is fast approaching.

Some recent AI posts on this forum are surprisingly legit looking. Some of these ask for help, some attempt to give advice. While sounding like a real person with something useful to discuss.

Even if their contributions are of some value, I'm not a fan. Because their end game is not in our best interest.

- Thomas

[TD-er]

I try to remove all of them when they appear to be spam.
Typically I remove about 10 accounts per week.
Feel free to report them if you encounter them.

[ThomasB]

I will send you a private email with the name of a recently joined member that is most likely AI.

- Thomas