Does using AI to write an OP break MN talk guidelines?

[CavalierApproach]

Asking because I’ve just been looking at a thread where the OP sounded really ChatGPT-ish to me — then as I was about to post with a question it was deleted, and the deletion message says it broke talk guidelines.

Just wondering if that might have been the reason?

I'd like to know about this as well.

There seem to be a lot of them these days, following three formats:

1. AI assisted opening post, just the opening post, OP disappears. The opening post is frequently lengthy, has shortish paragraphs, and a few 'hooks'.
2. As above, but there is one early but shorter follow-up response from the OP before they disappear.
3. An OP who has the time/motivation to keep the AI assistance running right throughout the thread.

As ever I think they're a mix of bored trolls who learned how to use AI assistance, lazy journalists who learned how to use AI assistance, and political stirrers.

They are bloody annoying 😡

The flats thread in money matters is still standing despite being obviously AI. Mumsnet don't seem to care.

Does AI make spelling mistakes? There are a few in the opening post of that flats thread

If the OP is using AI to assist them, they can alter the opening post to include 'an authentic mistake' or two, either manually or via instruction.

Hi @CavalierApproach we don't have a ban on people using AI to help them write their posts as such but we look at these things case-by-case. We have banned users who are quite obviously using AI to generate reams of text and are not really engaging properly with other MNers.

On the flip side, we've had reports that posts are AI when there's no evidence to suggest that's the case - it's just that person's writing style and we can see they've been that way since way before large language models became available to the masses!

Overall, we're not seeing a huge amount of this kind of thing outside of spam bots. As always, please report your suspicions to us and we'll happily take a look.

I have the impression that there are a fair number of opening posts that are started in order to generate responses to train AI - I really object to (mainly) women being used as free labour to train AI.

I thought that about the flats thread too, but then saw that the OP has posted comments on other threads which sounded human so wasn’t sure after that.

There was a really annoying, and long, AI post in an A’ Level / Uni choice thread recently. Everyone else was engaging with the flow of the discussion, the AI post completely lacked any nuance, specificity, or comprehension of the OP’s evolving thoughts across the thread or acknowledgement of other posters helpful contributions. It was just Every Banal Thing There Is To Say About University Choice. In numbered paragraphs.

I reported and it was deleted.

But as they improve and are inserted into more and more threads, and become harder to distinguish from real human interaction, it’s going to permanently and irrecoverably alter the nature and quality of discussion here.

One thing I’ve been wondering about is whether linked words in the text are a sign of AI. But it might also be something MNHQ are doing.

That's an excellent point.