Please or to access all these features

Site stuff

Join our Innovation Panel to try new features early and help make Mumsnet better.

See all MNHQ comments on this thread

Odd AI-generated thread title options in Site Stuff

205 replies

AmIAComplete · 04/10/2026 19:33

AIBU Response Schema v1.0: Structured adjudication for interpersonal reasonableness posts about family support during a child’s serious illness, with special handling for ND context and empathy calibration. Use this schema to produce concise, practical verdicts and communication guidance without escalating stigma or family conflict unnecessarily. The assistant should assess whether the poster is being unreasonable, acknowledge the child’s medical context, weigh the brother’s expectations and possible ND-related rigidity, and give a clear, forum-style ruling plus brief actionable advice. Definitions: 'AIBU' = 'Am I Being Unreasonable?'; 'ND' = neurodivergent/neurodivergence; 'DD' = dear daughter. Safety/quality constraints: do not diagnose anyone; do not demean ND people; do not imply a sick child owes social performance; do not overstate certainty about motives; distinguish the child nephew’s needs from the adult brother’s message; prioritize compassion, boundaries, and de-escalation. Output should be parseable JSON matching this schema exactly, with no extra keys, comments, or prose outside JSON. In verdict, choose one of the enumerated values. In confidence, give 0..1. In core<span class="underline">judgment, give a short forum-style judgment. In reasoning, briefly explain balancing factors, including that serious illness and treatment commonly affect energy/mood; the brother may be stressed but his complaint is misplaced; ND can explain disappointment but not obligate the ill child to perform. In advice, include immediate emotional validation and one or two practical next steps (e.g., do not engage tonight, send a calm boundary later, loop in parents only if needed). In suggested</span>reply, provide a short message the poster could send to the brother, neutral but firm, avoiding argument. In tone<span class="underline">flags, indicate whether the response should emphasize empathy, boundaries, de-escalation, and whether to avoid legal/medical advice. In risk</span>notes, note any caution about the poster’s acute stress and potential family fallout. minor<span class="underline">child</span>considerations should explicitly state that children’s wellbeing takes priority over adult expectations in family gatherings when a child is newly seriously ill. nd_sensitivity should acknowledge ND respectfully while rejecting its use as a trump card over another child’s health needs. Keep outputs concise but humane, roughly the level of a supportive forum reply translated into JSON fields.

Weird shit on AI title options

But the AI worked on this post

OP posts:
Error404FucksNotFound · Yesterday 12:04

So you are saying that openai completely independently and bizarely generated a remarkably specific AIBU-response schema with full instructions for a detailed response post instead of a thread title?

TemperanceWest · Yesterday 12:05

The tool works by sending your draft to OpenAI, which is meant to send back a single suggested title

I think this needs to be made crystal clear to posters.

DuncinToffee · Yesterday 12:06

JaneJeffer · Yesterday 11:56

There’s no need for an AI title generator @JustineMumsnet

I agree

It is very pushy with those suggestions as well the way starting a new thread is set up now

escapegoats · Yesterday 12:06

TemperanceWest · Yesterday 12:05

The tool works by sending your draft to OpenAI, which is meant to send back a single suggested title

I think this needs to be made crystal clear to posters.

So do I.

MrsBennetsPoorNervesAreBack · Yesterday 12:06

Error404FucksNotFound · Yesterday 12:04

So you are saying that openai completely independently and bizarely generated a remarkably specific AIBU-response schema with full instructions for a detailed response post instead of a thread title?

That would appear to be the substance of explanation number 3.

So, my summary so far:

Explanation 1: It looks complicated but it's just what we use to generate titles.

Explanation 2: @AmIAComplete did it.

Explanation 3: OpenAI did it.

JustineMumsnet · Yesterday 12:08

MrsBennetsPoorNervesAreBack · Yesterday 11:59

@JustineMumsnet , with respect, this is the third different explanation that we have been offered by MNHQ this morning.

Are your moderators, or those who are guiding them, not trained to investigate matters thoroughly before posting spurious explanations on the site which don't stand up to scrutiny?

Do you recognise that this kind of issue seriously damages trust in the Mumsnet brand, and that the dodgy attempts to explain it away merely undermine the credibility of the site even further?

Yes I certainly recognise that posting too quickly without really getting to the bottom of things can be a mistake (i've done it enough times myself!) but equally you'll know that not responding promptly often leads to wild speculation and people thinking the worst, particularly where Ai is concerned. So it's a bit of a balancing act. Nonetheless we weren't thorough enough this morning and I'm sorry about that. I want to assure people though it wasn't because we've anything to hide - more cock up rather than conspiracy!

Megifer · Yesterday 12:08

So in a nutshell the site request is "generate a short engaging thread title based on the content"

But what was received was "generate the reasoning behind an AIBU response in MN forum style complete with the parameters followed"?

OR the right request was received, but a completely different response was given that also happened to be the reasoning that the AI would follow?

Ive come across AI outputs being inaccurate but thats pretty strange.

Megifer · Yesterday 12:09

Error404FucksNotFound · Yesterday 12:04

So you are saying that openai completely independently and bizarely generated a remarkably specific AIBU-response schema with full instructions for a detailed response post instead of a thread title?

Or what she said 😂

KitschIntoKitchen · Yesterday 12:09

I wonder if this thread will be promoted on Facebook or the Daily Fail? Probably not.
Where’s a good investigative journalist when you need one?

Error404FucksNotFound · Yesterday 12:10

Is this AI stuff that MumsGPT that your new ceo has championed?

How Mumsnet harnessed AI to become a data and insights powerhouse

JaneJeffer · Yesterday 12:11

Megifer · Yesterday 12:08

So in a nutshell the site request is "generate a short engaging thread title based on the content"

But what was received was "generate the reasoning behind an AIBU response in MN forum style complete with the parameters followed"?

OR the right request was received, but a completely different response was given that also happened to be the reasoning that the AI would follow?

Ive come across AI outputs being inaccurate but thats pretty strange.

And this is why we don’t trust AI

BellesAndGraces · Yesterday 12:15

Sorry@JustineMumsnet, but this still doesn’t really make sense. The AI didn’t just return a random reply instead of a title. It returned something called “AIBU Response Schema v1.0”, complete with fields for verdict, reasoning, advice, suggested reply, tone suggestions etc, and detailed instructions telling the AI tool how to respond to an AIBU post.

Are we really being told that none of that exists anywhere in Mumsnet’s AI setup and OpenAI spontaneously invented an entire AIBU response schema when asked to suggest a thread title?

You may not be able to see inside OpenAI’s model, but presumably you can see exactly what Mumsnet sent to it and exactly what it returned.

Horses and zebras Justine!

Morecoffeeneedef · Yesterday 12:17

Sorry @JustineMumsnet but it is hard to give much credence to these explanations. First we were told that this code was just creating a title, which is manifestly nonsense; then there was a suggestion that OP or someone else 'had that code in their browser' which again sounds very unlikely; and now that Open AI, which has presumably been making title suggestions all the time that facility has existed, randomly returned code to create a reply to the thread. How could that happen, and why wasn't it immediately obvious to whoever investigated the issue, that it had happened?
It feels that we are being fobbed off, and I for one mind very much if Mumsnet is creating, or allowing to be created, responses to threads. It's not a small thing.

Bloodysquirellls · Yesterday 12:22

SwedishEdith · Yesterday 11:55

Ha ha but you were right.

I’m always right 😉

😂

MrsBennetsPoorNervesAreBack · Yesterday 12:22

JustineMumsnet · Yesterday 12:08

Yes I certainly recognise that posting too quickly without really getting to the bottom of things can be a mistake (i've done it enough times myself!) but equally you'll know that not responding promptly often leads to wild speculation and people thinking the worst, particularly where Ai is concerned. So it's a bit of a balancing act. Nonetheless we weren't thorough enough this morning and I'm sorry about that. I want to assure people though it wasn't because we've anything to hide - more cock up rather than conspiracy!

Surely the obvious thing to do in that scenario would be to post a holding response to say "we have no idea what's happening here but we're investigating and will get back with an explanation asap" rather than posting explanations which are self-evidently incorrect.

Yes, there would be speculation regardless, but how on earth are we supposed to trust anything that the moderators say in future if we know that they have previously posted made-up explanations which were subsequently acknowledged to be wrong?

This really is a massive cock up on MNHQ's part, because the dodgy explanations really make it look like MNHQ is trying to hide something.

And if the media pick this up, it isn't going to reflect well on the site at all.

BillieWiper · Yesterday 12:23

So if you don't use that guff (sorry I'm not technical!) to make fake responses where did it come from? Why would OpenAI compel itself to do/try to do something so specific if it's never been asked to do that before from your side? It's like it's appeared in the wrong place. Rather than just sprung out of thin air.

It sounds really weird but I just don't know anymore what's real and what isn't sometimes!

TemperanceWest · Yesterday 12:24

MrsBennetsPoorNervesAreBack · Yesterday 12:22

Surely the obvious thing to do in that scenario would be to post a holding response to say "we have no idea what's happening here but we're investigating and will get back with an explanation asap" rather than posting explanations which are self-evidently incorrect.

Yes, there would be speculation regardless, but how on earth are we supposed to trust anything that the moderators say in future if we know that they have previously posted made-up explanations which were subsequently acknowledged to be wrong?

This really is a massive cock up on MNHQ's part, because the dodgy explanations really make it look like MNHQ is trying to hide something.

And if the media pick this up, it isn't going to reflect well on the site at all.

Exactly.

JaneJeffer · Yesterday 12:25

A big boy did it and ran away.

escapegoats · Yesterday 12:25

JaneJeffer · Yesterday 12:25

A big boy did it and ran away.

Grin
escapegoats · Yesterday 12:26

Morecoffeeneedef · Yesterday 12:17

Sorry @JustineMumsnet but it is hard to give much credence to these explanations. First we were told that this code was just creating a title, which is manifestly nonsense; then there was a suggestion that OP or someone else 'had that code in their browser' which again sounds very unlikely; and now that Open AI, which has presumably been making title suggestions all the time that facility has existed, randomly returned code to create a reply to the thread. How could that happen, and why wasn't it immediately obvious to whoever investigated the issue, that it had happened?
It feels that we are being fobbed off, and I for one mind very much if Mumsnet is creating, or allowing to be created, responses to threads. It's not a small thing.

Edited

Agree. If I want to chat to an AI bot I will just post my question in Meta AI or Claude or whatever

Its meant to be Mumsnet not MumsandAInet

Bloodysquirellls · Yesterday 12:29

As an aside I think you should get rid of the ai title feature. If someone can’t write their own title then how can they write their post.

The title sets the tone of the thread. It’s part of the threads identity and character.

It’s important I feel that this is written by humans.

Megifer · Yesterday 12:31

Would a rogue integration call itself "v1.0"? That indicates there are other versions of response parameters.

BeigeCardiganOfDoom · Yesterday 12:32

Can you tell us the exact prompt/ request MN sent to OpenAI?

MrsBennetsPoorNervesAreBack · Yesterday 12:33

JustineMumsnet · Yesterday 12:08

Yes I certainly recognise that posting too quickly without really getting to the bottom of things can be a mistake (i've done it enough times myself!) but equally you'll know that not responding promptly often leads to wild speculation and people thinking the worst, particularly where Ai is concerned. So it's a bit of a balancing act. Nonetheless we weren't thorough enough this morning and I'm sorry about that. I want to assure people though it wasn't because we've anything to hide - more cock up rather than conspiracy!

The more I think about this response, the more it bothers me.

It seems to be saying that MNHQ prioritised posting something quickly (in order to try to shut down the inevitable speculation) over and above posting something that was actually true.

BellesAndGraces · Yesterday 13:05

MrsBennetsPoorNervesAreBack · Yesterday 12:33

The more I think about this response, the more it bothers me.

It seems to be saying that MNHQ prioritised posting something quickly (in order to try to shut down the inevitable speculation) over and above posting something that was actually true.

This bothers me too. So does the admission that data is being sent to OpenAI.

Swipe left for the next trending thread