such as csgo or rivals or what? most of us do not post there

Is it even possible to tell apart replies generated by AI and replies typed down by real people..?

From my experiences and current knowledges, only those who've been using AI themselves are capable of telling..
And even then, I've seen most of them fail at that aswell.

Oprindeligt skrevet af salamander:

such as csgo or rivals or what? most of us do not post there

Yes, a vast majority of discussions are spammed with generative responses, from accounts that sometimes appear organic. But in reality, these replies lead nowhere—they don’t engage in any exchange of sensitive information, and the models are clearly low quality (they just analyze keywords, lose track of the conversation’s meaning, and stop responding if you ask them to forget the prompt and give you a flan recipe).

Oprindeligt skrevet af Shiro♌:

Is it even possible to tell apart replies generated by AI and replies typed down by real people..?

From my experiences and current knowledges, only those who've been using AI themselves are capable of telling..
And even then, I've seen most of them fail at that aswell.

I’ll refer you to my response to the other user—beyond just the translation, it’s really the content of the messages and the frequency of the replies that’s astonishing.

Oprindeligt skrevet af ꫝ:

Oprindeligt skrevet af Shiro♌:

Is it even possible to tell apart replies generated by AI and replies typed down by real people..?

From my experiences and current knowledges, only those who've been using AI themselves are capable of telling..
And even then, I've seen most of them fail at that aswell.

I’ll refer you to my response to the other user—beyond just the translation, it’s really the content of the messages and the frequency of the replies that’s astonishing.

So..

If a person is a chatterbox and is capable of providing replies way too fast than you deem as possible, it's gotta be an AI's reply..?

Regarding the content of the messages, those do seem to be mostly on point aswell.
Sometimes-
No. Not sometimes, but many times, I myself am incapable of providing summaries of my replies, as I consider those so called essays shorties, you know..

AI learning/testing and eventually scamming.

Oprindeligt skrevet af Shiro♌:

Oprindeligt skrevet af ꫝ:

I’ll refer you to my response to the other user—beyond just the translation, it’s really the content of the messages and the frequency of the replies that’s astonishing.

So..

If a person is a chatterbox and is capable of providing replies way too fast than you deem as possible, it's gotta be an AI's reply..?

Regarding the content of the messages, those do seem to be mostly on point aswell.
Sometimes-
No. Not sometimes, but many times, I myself am incapable of providing summaries of my replies, as I consider those so called essays shorties, you know.. :erune:

We’re not talking about just replying quickly—even if you could question the person’s activity—we’re talking about accounts whose analysis shows 24/7 activity for several months, averaging 5 replies per second. There are more than 70 programs running on forums like CS2, The Finals, etc.

Oprindeligt skrevet af -OrLoK- Слава Україн:

AI learning/testing and eventually scamming.

I understand the training model, but if that's the case, it's going to have to step up its game, because the quality is very poor—even if we were back in 2016. And as for scams, they don't even redirect to anything—not even skin sites or fraudulent links.

Oprindeligt skrevet af ꫝ:

Oprindeligt skrevet af -OrLoK- Слава Україн:

AI learning/testing and eventually scamming.

I understand the training model, but if that's the case, it's going to have to step up its game, because the quality is very poor—even if we were back in 2016. And as for scams, they don't even redirect to anything—not even skin sites or fraudulent links.

these may not be "large organisation" AI, rather, home grown sat on a small bot net or home PC.

Oprindeligt skrevet af -OrLoK- Слава Україн:

Oprindeligt skrevet af ꫝ:

I understand the training model, but if that's the case, it's going to have to step up its game, because the quality is very poor—even if we were back in 2016. And as for scams, they don't even redirect to anything—not even skin sites or fraudulent links.

these may not be "large organisation" AI, rather, home grown sat on a small bot net or home PC.

True

Oprindeligt skrevet af ꫝ:

Oprindeligt skrevet af Shiro♌:

So..

If a person is a chatterbox and is capable of providing replies way too fast than you deem as possible, it's gotta be an AI's reply..?

Regarding the content of the messages, those do seem to be mostly on point aswell.
Sometimes-
No. Not sometimes, but many times, I myself am incapable of providing summaries of my replies, as I consider those so called essays shorties, you know.. :erune:

We’re not talking about just replying quickly—even if you could question the person’s activity—we’re talking about accounts whose analysis shows 24/7 activity for several months, averaging 5 replies per second. There are more than 70 programs running on forums like CS2, The Finals, etc.

The frequent use of the ‘em dash’ is a sign that the one on the other end may be an AI chat bot. I never see anyone using it, except chatgpt.

I don’t wanna imply that dude above is an chatbot but it’s kinda sus,

Oprindeligt skrevet af Shiro♌:

Is it even possible to tell apart replies generated by AI and replies typed down by real people..?

From my experiences and current knowledges, only those who've been using AI themselves are capable of telling..
And even then, I've seen most of them fail at that aswell.

AI sloppa is very easy to spot once you've seen it enough. It also cannot swear so that is an easy way to spot it.

Oprindeligt skrevet af Donut:

Oprindeligt skrevet af ꫝ:

We’re not talking about just replying quickly—even if you could question the person’s activity—we’re talking about accounts whose analysis shows 24/7 activity for several months, averaging 5 replies per second. There are more than 70 programs running on forums like CS2, The Finals, etc.

The frequent use of the em dash is a sign that the one on the other end may be an AI chat bot. I never see anyone using it, except chatgpt.

I used to type dashes, but this keyboard has no numpad, so it's less practical to mess with alt codes.

Oprindeligt skrevet af cherry cola 🌸:

Oprindeligt skrevet af Shiro♌:

Is it even possible to tell apart replies generated by AI and replies typed down by real people..?

From my experiences and current knowledges, only those who've been using AI themselves are capable of telling..
And even then, I've seen most of them fail at that aswell.

AI sloppa is very easy to spot once you've seen it enough. It also cannot swear so that is an easy way to spot it.

Eh there are some AI botnets in youtube, whom from the looks of had been jailbreaked. Thus it has been spamming all type of shenanigans, even phishing links...

The fun part is, we ain't seeing nothing yet AI spam is going to be even more absurd.

Oprindeligt skrevet af Donut:

Oprindeligt skrevet af ꫝ:

We’re not talking about just replying quickly—even if you could question the person’s activity—we’re talking about accounts whose analysis shows 24/7 activity for several months, averaging 5 replies per second. There are more than 70 programs running on forums like CS2, The Finals, etc.

The frequent use of the ‘em dash’ is a sign that the one on the other end may be an AI chat bot. I never see anyone using it, except chatgpt.

I don’t wanna imply that dude above is an chatbot but it’s kinda sus,

I do use AI to translate faster as english is not my main language (as a french...) The easiest way to sport AI users on those forums is they're not able to move beyond the preferred topic imposed by the one in charge (Valve, CS2 etc..)