Ten fold

« first day (5261 days earlier) ← previous day next day → last day (42 days later) »

Stephan Kolassa

09:16

Here is a new wrinkle: people post rather trivial (in the original sense of the term) homework questions, and self-answer. stats.stackexchange.com/q/659857/1352 and stats.stackexchange.com/q/659855/1352 and stats.stackexchange.com/q/659852/1352. Now that they are answered, they won't even be deleted if they get closed. Let's hope this is a one-time occurrence.

Sextus Empiricus

09:43

@StephanKolassa I had the idea that there recently were a lot of trivial questions written in some confusing way and aome self answer in the text with the question 'is this correct?'. It made me wonder whether the people that are playing with AI had turned from confusing answers to simplistic questions. Then I saw this post and thought, there's another one stats.stackexchange.com/questions/306957 but it just happened to be a bumped up post.

Actually, they look very much like AI's (or at least very useless answers) if you look through their answers and questions. The old question that I came across and was bumped, is actually bumped due to a (useless) answer by the same user that you noted.

Stephan Kolassa

10:23

@SextusEmpiricus To be honest, these do not look like AI to me. They are too unsophisticated. It more looks like a couple of students recently learned about the existence of CV and misunderstood us as their homework tutors. Could of course also be a bunch of sock puppets.

In any case, I have flagged one post and hope the mods can do something about these "users".

Gah! These sock puppets are upvoting each other faster than I can downvote them! (I expect I'll next be auto-flagged and auto-suspended for serial voting...)

User1865345

10:48

@StephanKolassa yes. These are sock puppet trolls and not some gen ai users. Troll accounts have of late proliferated across SE networks.

And no you will not be auto suspended @StephanKolassa 😅

Sextus Empiricus

11:03

@StephanKolassa It is only a theory for the moment. I had been thinking about it because of the recent AI's answering in the typical verbose chatGPT style, and after that got sushed it seemed like there came all kinds of new users with weird questions instead of answers.

Now this new case is not just a student. They are not just asking and are also answering in several old probability related posts with nonsense answers stats.stackexchange.com/a/659862 there is likely something dishonest going on there.

For me I have a high prior that this is AI training. A lot of people nowadays get paid to train AIs, why not go to stackexchange and ask the AI a prompt: "create me an interesting answer to this question in the style of user ... " and then use votes or responses as input for the AI training. Let the bot chat with random users on the internet for training.

User1865345

@SextusEmpiricus i had no idea this is how they are trained.

Sextus Empiricus

@User1865345 I get headhunted for this on LinkedIn, to train language learning models (sadly it is the only thing that I get headhunted for).

User1865345

😅

If this is the way, this seems so dumb.

The gen ai is apparently way developed to compose authoritative looking answers, as we have noticed recently.

Sextus Empiricus

I sometimes wonder whether that other recent AI got to answer questions in the style of user1865. All that stuff about measure theory.

User1865345

Lol. I never thought my answers had any commonalities or structure.

But one thing for sure. Those AI answers are pretty solid and often not that nonsensical.

Even though few are outright gibberish.

But I wasn't aware of AI training pov.

Doesn't SE already provide materials for AI training? Pardon me for I am not familiar with all that things happening.

Update: one user has been suspended.

This is outright trolling.

I have removed the answer url. Looks odd.

Stephan Kolassa

11:54

OK, it seems like I just deleted the last Q. Let's hope for a little more peace and quiet here.

@SextusEmpiricus Now I'm disappointed. Is someone training an AI to answer in my style? (Given many of my recent interactions here, I would assume I could definitely be replaced by an AI... sometimes it looks like I'm repackaging the same five answers for slight variations of the ever-same questions that just vary a tiny bit so I can't dupe-close them...)

Dear moderators, can you please dispose of "users" Acoustic Fencing Panels Ltd, Addiction Rehab Clinics Ltd, Store Frontsltd and User Call Handling Ltd? Thank you. stats.stackexchange.com/users/457554/… and stats.stackexchange.com/users/457575/… and stats.stackexchange.com/users/457573/store-frontsltd and stats.stackexchange.com/users/457570/call-handling-ltd.

Stephan Kolassa

12:09

I have posted a request to be able to flag entire users at Meta.SE, meta.stackexchange.com/q/405569/256777, in case anyone is interested.

Stephan Kolassa

12:21

And while we are at it, there is also Tennis Court Resurfacing Ltd, stats.stackexchange.com/users/457529/…. It looks like the mods or automated tools have been quite successful in pruning these back, because searching for "Ltd" among the users does not yield any other obvious spammers.

User1865345

12:32

@StephanKolassa great. Upvoted.

I appreciate our vigil community for being alert and raising some good proposal. 👏🏻

That suspended user has now started shit posting at English learners.

Profile banning feature is really appreciated. 🙂

Dec 24, 2024 at 14:50, by User1865345

Q: Can you please consider multi site behavior in automatic bans?

If I'm correctly understanding this answer, behavior and bans on one site do not automatically affect a user on other SE sites. For the past few weeks, what appears to be a single person has been posting hundreds of unserious or nonsense posts with multiple sock puppet accounts. This individual a...