To be suspicious about these very similar posts

This reads so much like so many posts on here at the moment 😂

I think, even before AI Mumsnet had plants posting as genuine users, not that they’d admit it.

I also swear that they just repost old threads sometimes but with an update date. Like I can remember the initial posts and all of the subsequent replies!

i am sure i read an thread here, and it was later in a paper, of some sort,
favourite bbc personalities, lots of similarities including Jane somebody, Danny Dyer
weird

Wow, that AI-generated post is right on the money!!

Yep, there seems to be more and more AI generated posts as well as posts that are clearly made up, possibly by a human being. Then there's all the rude nasty replies when someone posts about a perfectly reasonable topic. Clearly meant to create discussion.
My personal strong dislikes are all the adverts for clothes and beauty products masquerading as genuine queries eg 'What do you think of this dress as a wedding guest outfit" or 'Did you know that a particular shop now sells xyz, it's really helped my skin' etc.

So I asked it to generate a typical Mumsnet post which I’ll paste below. It’s got a lot of typical tells, but bear in mind it’s unedited and a lot of people who use AI will edit it a little to try and make it more obvious:

Subject: AIBU to be annoyed about DH’s idea of “babysitting” his own kids?

Hi all,

Bit of a rant incoming – sorry in advance!

So DH (darling husband – although “darling” is doing a LOT of heavy lifting right now) has just texted me at work saying he’s “babysitting” the kids tonight so I can go to my friend’s birthday dinner. Which, on the face of it, sounds nice… except THEY’RE HIS CHILDREN TOO.

I mean, how is it babysitting when it’s literally just… parenting? He acts like he deserves a medal every time he’s left alone with them for more than an hour. Meanwhile, I’m expected to juggle work, school runs, packed lunches, laundry, and organising a birthday party for 25 sugar-fuelled six-year-olds while also pretending to enjoy the 17 WhatsApp group chats I’ve somehow ended up in.

Also, let’s be real – by “babysitting” he means letting them stay up late eating Wotsits while he watches Match of the Day and pretends he can’t hear the three-year-old shouting that she’s weed on the carpet again.

AIBU to find this infuriating?

(Also, just to add – he’s a decent dad most of the time and I do love him, but this mindset is driving me round the bend.)

None of these things are “proof” in-and-of-themselves, but the typical hallmarks of AI-generated text (particularly ChatGPT) are:

1. em-dashes—e.g., a double dash, often without a space before and after, like this.
2. US spellings.
3. Tripartite lists (I’m crossing my fingers, legs, and eyes).
4. Oxford comma (example above)
5. Very structured text - lots of headed paragraphs, particularly when the subheaders are bold.
6. Concluding sentences at the end of most paragraphs.
7. Might refer to “experts” or “studies” without specifying what they were, though this is also so common with humans that it might not belong in this list!
8. Often follows a rigid format: introduction, three body points, conclusion — even when not appropriate.
9. Each paragraph may be the same length or rhythmically similar.

I used AI to generate the last two points. 😅

Awesome, thanks!

Ah, there was a religious one that had the same issues. They were also able to write really long replies really fast, which was another tell.

Cube of poo?! Heck that sounds like a good one and I've clearly missed it!

Lots of em dashes is a good one. As is overly “neutral” language.

Lots of fillers.
Starting with phrases like “okay so” or let’s be honest.”

Repetition and filler.

Lack of personality, but also too much personality that doesn’t seem authentic.

But really, it’s just that if you read a lot of it, you can just tell. It all sounds very same-y in a way I find hard to pinpoint but can pretty regularly spot!

If you read People.com you'll find a lot of stories lifted from Mumsnet

Yes! Or ‘cube of poo’ or ‘tiny Korean woman’ 😂

It wasn’t religious. The layout was so formulaic, particularly the layout of the various rebuttals.

They're suggested names when joining. Newspapers are using that system too.

We need an AI-spotters checklist! 😀

That’s definitely weird!

Unless it's a very new post with few replies then I look to see if the OP has a "see next" on their first post. If it's several pages and OP hasn't returned then I don't even bother reading it.

Well spotted. Definitely.

It's generated content to drive traffic to the site. If it becomes popular it will then appear on the Mumsnet Facebook page, and might get picked up as a 'news' story which then brings more traffic to the site and increases ad revenue.

If it's free, you're the product...

Was it a post with a religious theme? If so it was blatantly AI, but then tons of posts on here are blatantly AI these days. As someone who works with AI daily, I can spot them a mile off and I have to sit on my hands not to say anything.

I'm starting to think the only posts worth replying to are ones where the OP has either posted a few times on the tread already, or showed some other sign of being a regular. Bring back the days of stating "cutted up pear, sistine chapel, penis beaker" to establish one's credentials! 😀

The thing giving me the suss these days is an OP posting some issue they want advice on and then never coming back. Journalists trawling for stories vibes! Seem to be loads of them recently!