New rule against AI use on ES

ps, i have a social media account i polled people on.
Question was, do you navigate away from websites that look like they're full of LLM output that's unedited?

Sample size: 32
93.8% said yes
6.3% said no

Sounds like i'm not alone in wanting to read robot spew
 
Last edited:
I'm glad you switched to LLM. Having done a fair amount of Artificial Intelligence work in the 90s, the use of the acronym "AI" these days gets my blood pressure up. IMO, AI these days is a PR conduit for companies like NVIDIA to up their stock price. Everyone gloms on the word based on what they see and hear and...bingo it's a thing to worry about. Kind of like "swarming UAVs" and "intelligent agents". My own personal criteria for artificial intelligence is that it has to have the capability to modify its behavior based on external stimuli of unforeseen events. It has the ability to gain knowledge and operate within constructs that were never originally envisioned at inception...similar to what was presented in Terminator. In this behavior, it is also subjected to viruses much in the same manner as humans where it's behavior can be corrupted through a number of false or surreptitious learning events. This currently does not exist in its entirety with what is know as AI today.
Maybe you could help make my wolfybot do that. :)
 
I wasn't sure about that one
 
Their profile was full of spam links (removed those too), their email is for a spammer, (is on stopforum spam) as well as their ip. So...was an easy call.

Whether or not it was actually ai spam or not I dunno, but it reads like it.

Interestng, i still had their IPs used page up and it refreshed itself whe I went back to it, and it ahd a new Ip it said they accessed from *after* I'd banned them and edited out their seospam in username and profile, and that IP is even *more* used by spammers per stopforumspam. :roll:

EDIT: Now they created a new account to spam from, totally new IP, determined little ****s. :/
 
Last edited:
I noticed something odd over the last couple of weeks. It appears to happen only when using my iPad, but could be happening with my iPhone as well. It's as though AI has taken over my keypad. I couldn't figure it out at first. I'd post something, and look at it after posting, and it wasn't what I had typed, so I'd need to go back and edit the post. I started paying more attention, and similar to spell check, my sentences were being changed, added to, or completed automatically. It was weird, because the words it added had ebike terminology, so at first glance looked sort of right, but definitely not after reading it. Then watching carefully while typing, I could see up to 5 or 6 words being added after my last typed word. Anyone else notice this?
Autocorrect is bad enough when it's wrong, but this is really annoying.
 
I don't use any of the technology that does that sort of thing, including autosuggest / spell / etc, because they are ALWAYS wrong and just screw up everything I type.

If necessary I hack that out of something if it's in a program I have to use and doesn't have a simple "no" button to permanently disable it. If it isn't hackable (by deleting files or registry entries or services) then that entire program goes away off the machine or device. ;)
 
I am with some of the above this post. I have no idea how to detect AI anything and really don't care to waist my time doing so. I would rather play my guitars, weed my garden and grow good food to eat, ride my bike and wish you all the best of luck.

However is ES want's to post some sort of class or video (youtube style) on how to detect AI, I will be happy to view it.

:)
⚡
 
I noticed something odd over the last couple of weeks. It appears to happen only when using my iPad, but could be happening with my iPhone as well. It's as though AI has taken over my keypad. I couldn't figure it out at first. I'd post something, and look at it after posting, and it wasn't what I had typed, so I'd need to go back and edit the post. I started paying more attention, and similar to spell check, my sentences were being changed, added to, or completed automatically. It was weird, because the words it added had ebike terminology, so at first glance looked sort of right, but definitely not after reading it. Then watching carefully while typing, I could see up to 5 or 6 words being added after my last typed word. Anyone else notice this?
Autocorrect is bad enough when it's wrong, but this is really annoying.
Samsung has also added "artificial intelligence" to their user interfaces. I haven't seen it do that yet though. Try a different keyboard app, or digging deep into the settings of whatever one you're using.
 
Samsung has also added "artificial intelligence" to their user interfaces. I haven't seen it do that yet though. Try a different keyboard app, or digging deep into the settings of whatever one you're using.
Google AI says this about the predictive text used by IOS. I'm going to turn the feature off. Seems like more trouble than it's worth and dangerous if you're not paying attention.

  • AI-Powered Suggestions:
    iOS uses a machine learning model, specifically a "transformer model," to analyze your typing patterns and predict the next words or phrases you might type.

  • Inline Predictions:
    You'll see suggestions appear as you type, either as inline predictions (gray text) or above the keyboard.

 
It's as though AI has taken over my keypad...Anyone else notice this?
I generally find the predictive word suggestions mostly helpful (but not always, by far) yet often annoyingly stubborn and incorrect. It seems to have learned my common phrases, which is the most helpful part.

Siri is turned off, and I haven't noticed any more sophisticated AI-level takeover like you described. But I will study it more closely now. Interesting.
 
We are probably going to have to build some specialized tools to find them automatically because it's becoming more common.
Riffing off the "SEO" statement, SEO Optimizers "backlink" their spam, in order to increase and then forward their "juice" to their intended URL. Meaning that Forum Staff could run a search of a forum post, in order to see if it's backlinked, and also carefully monitor which accounts have dropped a link to, say for example an Amazon listing. Any and all HTML links should be scrutinized carefully, since that is the ultimate objective of all spammers.

Which is not to say that all HTML links should be banned, however the spammers are dropping them to make their money; the Forum has a right to charge for the privilege. Even a very small and nominal charge will drive off 99% of all spammers, if in the event that Forum Staff wants to remain "friendly" to well-known and "local" members.

I distrust any and all AI. I distrust the acronym "AI", and resist it's use. It's not "intelligent" until it becomes sentient, and to ascribe a computer program that garbage outputs whatever garbage is inputted as "sentient" is extremely dangerous. Further, ascribing the "source" to AI shields the entity (person, corporation, government) behind the language (sales text, political propaganda, disinformation, etc...) from scrutiny. Meaning that an intelligent inquiry into WHY the "AI" decided to output x, y or z gets stymied because it's "AI", and who-knows-why this sentient program said "X".

Except it's NOT sentient, and some programmer is behind it all, and for that programmer's reasons.

And it's not about whether or not 1 out of 8 reasons to be concerned about battery fires are listed, it's why only 1 our of 8 reasons to foment war between Russia and Ukraine are talked about., Etc...

The comments about the Butlerian Jihad (Dune) were spot-on, and extreme and draconian means to kill any attempt at using AI will be necessary to prevent the enslavement of all humanity for all eternity. We are now in a pre-war condition for our survival, IMO.

Also, as the current political landscape stands, the one way to be 100% certain that a post is not AI, is whether or not it posts "racist" or "anti-Semitic" content. That should give some intelligent minds something to think about, when attempting to discern who might be behind the programming of the not-sentient, not-intelligent, GIGO "AI", and for what purposes it censors certain kinds of speech, while over-promoting disinformation and lies.
 
Not really on the thread topic, but:

Riffing off the "SEO" statement, SEO Optimizers "backlink" their spam, in order to increase and then forward their "juice" to their intended URL. Meaning that Forum Staff could run a search of a forum post, in order to see if it's backlinked, and also carefully monitor which accounts have dropped a link to, say for example an Amazon listing. Any and all HTML links should be scrutinized carefully, since that is the ultimate objective of all spammers.
SEO doesn't need links; it just has to contain the words that suggest or name the thing being optimized. Most of the SEO here is members that sign up as the name of the thing or place they're spamming for. A lof of the rest just put the name in their profile somewhere, and/or have their profile image or banner as the name / logo of their spam source.

Almost none actually post anything at all (the ones that do are very easily caught and filtered out with almost no effort; the ones that do it the other way require someone (me, mostly) to go thru each profile and actually see the page to find them, then remove their spam one profile at a time (because there is no way to do it en-masse).

There are also some "retarded" spambots that dont' stick their spam in the visible fields, but instead just have email addresses that are the spam. (some of them aren't even real addresses so the profile can never be validated so they could never log in and actually post spam).

There are *many many many* more "retarded" spambots that simply create thousands of randomly-named accounts that are never used for anything and contain no profile info. Many of these also use random garbage email addresses and so cannot even be activated as they'll never receive the activation email. Sometimes we're lucky and they'll use a domain or other definable term that is not used by real accounts and so can be filtered out to prevent some of them from signing up at all, until their bot changes to some other domain or term.

Some of the random junk accounts are "latent" spammers, which are handed off to other bots (or human criminals) that then pollute the forum with their profile spam. This is even harder to detect and remove because it happens long after the processing during signup, so the only way to catch it is to just continuously waste hours going thru all the thousands of profiles on the forum and see if anyone has added spam to it, then remove it one profile at a time.



Which is not to say that all HTML links should be banned, however the spammers are dropping them to make their money; the Forum has a right to charge for the privilege. Even a very small and nominal charge will drive off 99% of all spammers, if in the event that Forum Staff wants to remain "friendly" to well-known and "local" members.
Charging for advertising would do absolutely zero nothing zilch about the spammers.

Spammers do what they do completely separately from anything else, just like other criminal elements of society, to gain for themselves by using other people's resources for their own purposes without any form of recompense, just because they can.

I'd guess that keeping this forum clean of spammers takes at least 10-20 hours per week, between me and the rest of the mod/admin team, which is all unpaid wasted time we could be doing real useful things like helping more real users, organizing the data we already have so users can find it better, and working on our own projects.

If I couldn't pre-filter anything it'd probably be a lot more than that; it's getting harder to be able to create useful filters as spammers change methods.



Even without any form of Ai assisting the process, and just human criminals using simple bots and/or other human criminals to create the spam, eventually the internet will be all spam and no useful content, by ratio. Some places have already approached or exceeded more spam than content. Some forums and other resources have actually gone private pay-to-access because the problems were so bad that the companies or groups running them couldn't afford to control the spam anymore. Some simply shutdown and disappeared.
 
RE: Spammers

We're going to build some advanced anti-spammer tools soon. Spammers of today are tricky and exploit a lack of visibility in this software. What exists for Xenforo as plugins are inadequate. AI can now operate web browsers and circumvent bot protection mechanisms. It's probably true that recent advanced tactics stem from this.

I have an advanced tool ~80% designed, and it is the next major thing to build.

RE: AI detection

I wish AI detection was as easy as noticing that a user posted something racist/nationalistic/politically agitating; someone who could solve the problem has a billion dollar business on their hands. I can promise you that the solution is not that easy though. And most of that content is from Real "Intelligence" :D

We will never charge for posting. We will use advanced technology instead, if we need to fight AI with AI, we will. And we will equip fellow forum operators with the same weapons to keep the ecosystem alive.

agi-meme.png
 
Back
Top