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

384 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:
Ereshkigalangcleg · 12/11/2026 12:19

Is Perplexity a separate AI? I thought it just amalgamated all the different AIs, but I suppose the distinction is a bit meta! Maybe I’ve not understood what it is properly anyway!

C8H10N4O2 · 12/11/2026 12:54

Ereshkigalangcleg · 12/11/2026 12:19

Is Perplexity a separate AI? I thought it just amalgamated all the different AIs, but I suppose the distinction is a bit meta! Maybe I’ve not understood what it is properly anyway!

Its a separate company but its tools can integrate with the pro versions of ChatGPT, Gemini and Claude or use its own models. Initially it was the small relation but with Amazon backing and recommendation engines as an area of focus it could well end up specialising in that space.

VimesandhisCardboardBoots · 12/11/2026 15:28

MrsBennetsPoorNervesAreBack · 12/11/2026 12:07

Totally agree. Nobody should assume that the stuff that they're posting is totally anonymous.

ChatGPT didn't throw up massively concerning results for me, but that's not to say that it might not throw out more accurate results for others. Or indeed, that it wouldn't produce more accurate info about me with a slightly different prompt.

None of us should assume that we are unidentifiable.

As with anything involving AI, results may vary based on the whims of the probability gods.

I got the above with the free version of ChatGPT. I think it got fairly lucky with the threads it randomly sampled that time, and spectacularly lucky with the whole username connection, because the chances of it finding anything based on a Discowrld connection were tiny.

When I tried to replicate the results in another conversation still on ChatGPT, it did far less well, not really finding any threads with identifying info at all.

The only other LLM I tried was Grok, and that didn't seem to be able to search Mumsnet at all, so it may be that that one actually respects the sites requests to deny AI scraping. Which surprises me given it belongs to Elon Musk!

Ereshkigalangcleg · 12/11/2026 19:15

C8H10N4O2 · 12/11/2026 12:54

Its a separate company but its tools can integrate with the pro versions of ChatGPT, Gemini and Claude or use its own models. Initially it was the small relation but with Amazon backing and recommendation engines as an area of focus it could well end up specialising in that space.

Cool, I might look into it more then! I’ve got a job interview coming up and I need to talk about how best to use AI.

JaneJeffer · 12/11/2026 19:39

So it seems AI is less efficient at deducing someone is the same person than one’s own brain. There are people who post in completely different styles depending on username but I still know it’s them 🧐

Ereshkigalangcleg · 12/11/2026 19:40

JaneJeffer · 12/11/2026 19:39

So it seems AI is less efficient at deducing someone is the same person than one’s own brain. There are people who post in completely different styles depending on username but I still know it’s them 🧐

Same here.

C8H10N4O2 · 12/11/2026 19:46

Ereshkigalangcleg · 12/11/2026 19:15

Cool, I might look into it more then! I’ve got a job interview coming up and I need to talk about how best to use AI.

The answer is always "the tool which best fits the job" where the job is achieving a business benefit rather than an end it itself. Or the tool which is the best overall fit for a company's usecases if only one is desired. Look for usage examples in your industry/role area and what is used.

Its very good at some tasks, not worth it in others - knowing the difference is key to seeing a business benefit rather than simply something for a CTO to put on their CV.

The kind of fuzzy matching done on large volumes of data such as with recommendation engines, fraud detection, analysis of complex bundles of documents are popular both commercially and in the public sector. Along with the office assistant type tasks they are probably amongst the best known by business users. The automation coupled with a degree of decision making is probably most popular with tech audiences and for cutting costs and risk in service availability. That and the cyber protection potential.

C8H10N4O2 · 12/11/2026 19:49

JaneJeffer · 12/11/2026 19:39

So it seems AI is less efficient at deducing someone is the same person than one’s own brain. There are people who post in completely different styles depending on username but I still know it’s them 🧐

I found the paid versions could spot that (and more), the free versions less so but I'd guess that is the lower capacity limits in force. Some syntax and habit specifics are more reliable signals that the content of posts as they are near impossible to disguise. I still reckon I can spot this more reliably but I've been reading these posts for too long.

Ereshkigalangcleg · 12/11/2026 19:52

Yes I said similar to the bit about choosing the right tool for the job in my draft answer (practicing interview with Claude as I’m so rusty!) - will have a think about differentiating between them for different tasks.

New posts on this thread. Refresh page