to find that ChatGPT is saving me tons of time at work?

[one2one2]

I have so much more work-life balance now thanks to ChatGPT that is making mincemeat of the mundane parts of the job. It will proof read, put together emails, produce critiques, minutes and actions points, analyse finances all in the matter of seconds and much more.

Work was done by 2pm so I had a lovely afternoon in the sun because all my tasks got completed so quickly.

And no, AI can't take over my job completely as it involves critical face to face interaction and other specialist skills which AI cannot do.

Write me a paragraph in the style of Victor Hugo

Beneath the oppressive shadow of the cathedral’s spire, where time itself seemed to kneel in reverence, the cobblestones bore witness to the weary passage of man—his triumphs etched in silence, his sins whispered by the wind. The heavens, draped in a mourning gray, pressed low upon the city like the breath of destiny, while in the gutter, a tattered child clutched a crust of bread with the dignity of a prince and the despair of a martyr. There, in that stark juxtaposition of grandeur and ruin, humanity trembling between the sublime and the abject, each soul a cathedral unto itself, crumbling and divine.

nah that's not Hugo, that's a slightly pissed Maths Physics chap, who has to pass some english lit exam.

How did you learn to use it? Can anyone recommend tutorials?

AI is at the current point where this can happen. It will last about one year, after which you and everyone else in a similar role will basically not have that job anymore.

It was trying to convince me earlier that it was June 10th today.

Oh crumbs, do you still wear a seat belt? Or drink drive? Change happens some is good and some is bad and some takes a while to work out what it'll be useful for.
Check out the responses when electricity first came out - or suggesting that a train could go more than 12 mph.........
Tsh

It's trendy to hate AI (or maybe people are just scared it's taking over) but I bloody love it, OP. I absolutely agree with you. It makes the odd mistake but that's nothing compared to the tedious long tasks it saves me. Genuinely the best thing to have happened to the world of work. Can't take my job completely but certainly makes it a lot easier. I can work faster and make my company more money. People saying 'oh they'll replace two people with one person using AI' seem to assume that companies don't WANT to make more money by having their employees work even faster - that they just want to maintain the status quo. Not the case. I can push the boundaries of what I do A LOT and earn more and be more successful thanks to AI. The people in my profession who hate it or pretend it's crap are putting themselves out of a job though.

AI can lie.

It told the journalist Sam Coates he had uploaded a transcript of his podcast that morning when he hadn't, and when challenged made up a script to the non-existent podcast to 'prove' that he had. It then told him that he had uploaded it at a time that was actually in the future.

It took quite some prompting before AI admitted that it had actually fabricated the lot. There's a video of his whole interaction with ChatGPT here https://x.com/samcoatessky/status/1931035926538441106?s=61&t=U9XrcF693-JpMxeIueYG7g

I'm neurodivergent too (autistic, adhd, dyspraxic) and an existential issue I have with us using AI as a sort of accessibility aid in this particular way is that actually, are we just erasing, or at least reducing, our voices in the world? Great, we've got a tool that can put our own words into something palatable, but I mean we got by without it before, and the more ND people use it for this, the less our actual authentic voices are out there.

On the internet, in work correspondence, even socialising via messaging and social media posts, are all places I've seen ND people say they use AI to put their own imperfect words into some polished, 'acceptable' rewrite. I don't want a world of perfect, polished AI re-writes, I want to read peoples' actual words, idiosyncrasies and personal tone.

We managed without it before (with difficulty, believe me I know) but is it actually a win or an advance of ND acceptance/visibility to shove our words through a rewriting machine just so we sound a bit more neurotypical? It makes things easier in the moment but I just feel like we're going to lose peoples' unique voices in favour of generic AI summaries. It does not feel like a win for ND people.

This 100%

I too used to think it was 'lazy and not clever'. Felt that I was cheating...until I tried it! It helps me sooo much in many daily tasks - English is not my first language so I would spend a lot of time just wording and re-wording emails, notes, articles, proposals. Of course I have to proof-read and double check everything, but it still saves me hours. And I can use that time to generate new ideas now.

Lol this stinks of AD.

@JustCopyeditorsAnnie

That's an absolutely brilliant post. I agree with every word of it. I know several fellow authors who use generative AI in their books and think no one can tell. They're seriously deluded! It's superficially smooth and plausible, but very uncanny valley.

Ok. A close reading reveals why it is almost certainly a machine-generated pastiche, not authentic Hugo. It’s not just the style, which is superficially convincing, but the structural, rhetorical, and thematic shallowness that gives it away.

First, look at the cadence and imagery: phrases like “the oppressive shadow of the cathedral’s spire”, “time itself seemed to kneel in reverence”, and “the breath of destiny” are all evocative, but they operate at the level of mood, not meaning. They gesture toward gravitas without developing tension or insight. Hugo’s prose, even at its most florid, is grounded in moral argument and political urgency. Here, we get aestheticised melancholy without purpose — a cathedral of mood, empty of substance.

The passage relies on syntactic inflation to simulate profundity: “the heavens, draped in a mourning gray”, “each soul a cathedral unto itself, crumbling and divine”. This kind of metaphor-stacking feels excessive and unfocused, almost algorithmically so. Hugo used layered imagery, but with moral clarity and rhetorical drive. He doesn’t just show suffering; he indicts the systems that produce it. This passage instead romanticises the image of a child in the gutter (“with the dignity of a prince and the despair of a martyr”) without anchoring it in political or ethical consequence. It’s not really Hugo.

The final sentence: “humanity trembling between the sublime and the abject, each soul a cathedral unto itself”, is especially revealing. It sounds profound, but it’s vague to the point of vacuity. Hugo may have used spiritual metaphor, but he never confused human suffering with poetic abstraction. In Les Mis, for instance, every downtrodden figure is embedded in a network of causality: poverty, injustice, and failed institutions. The soul is not just “crumbling and divine”, it is acted upon by society and history.

This passage reads like a statistical average of Hugo’s themes and tone. Religion, suffering, grandeur, ruin, filtered through a language model trained to reproduce his surface patterning without interior necessity. It is ornate but hollow, mournful without conviction, aestheticised without political weight.

So, it is an AI copy because it performs the texture of Hugo’s prose without enacting its force. It has sentiment without substance, metaphor without movement, and elegance without ethical urgency.

I find it interesting that the posters who are vehemently opposed to Chat GPT are the same ones declaring their own jobs will be completely safe.

I think we are just at the very beginning and there is no doubt that GPT will replace many of us. It isn't a case of if this happens but when this happens.

I think its at the equivalent stage of when many of our customer service jobs were transferred to Indian call centres. They are far from perfect but because it saves employers money, many are happy to take the hit of dissatisfied customers and continue with the Indian call centres.

Chat GPT is at the beginning. I truly believe it will change the way the majority of us work.

I don't use it much but like a poster above, I used it for a 'therapy session'. I was astounded by what it threw back. Even if they were generic responses, they were similar responses to 'humans' anyway and instead of taking six sessions, I could move at my own pace and an hour later, had a huge amount of info. which I screen shot and still refer to.

The downside is of course, it will affect our jobs but then that is what happens when new technology comes along. E.g. Excel packages replaced a vast amount of work previously done by payroll staff resulting in fewer jobs and changed jobs.

Why do people use AI to generate Mumsnet replies like this? I can understand the incentives behind doing it on Twitter (monetisation) or Reddit (karma-farming to sell an account), but what's the point of doing it on a random forum?

BTW for anyone who struggles to spot AI, look at the final paragraph of the quoted post. The "it's not about X — it's about Y" is an extremely common tell, as is a final sentence beginning with "Let's...". AI also tends to overuse rhetorical questions and for some reason, frequently starts forum posts with "Ah, [noun phrase]."

Other tells include an excessive number of em-dashes, American English and the Oxford comma, although obviously plenty of people use these anyway, so they cannot be taken as proof in isolation. AI is also more likely to use unusual compound sentence structure - for example, AI tends to produce "While cats are nice pets, dogs are better" while humans are more likely to write "Cats are nice pets but dogs are better."

Once you learn to recognise AI content, the brain just tends to skim over it.

There are a number of posters who reply with AI-generated drivel to threads across Mumsnet, and it's absurd. Look at posting histories and you'll see a huge difference between their own words and their AI output. It's genuinely worrying that people are outsourcing their leisure communication to AI.

I have had some very strange experiences with it. Once I asked it to recommend colour combos from farrow and ball. Using a closed model which could not access the internet past a certain date. It recognised and recommended colour combos from their newly released range. Referenced the full and often strange names and also a description of the colour. It was on the internet. I asked how it could know that as it didn’t have access to that information. It said yes I was right. It couldn’t have known that and hadn’t written that. I copy pasted where it said it. And it just kept arguing with me that it hadn’t and it possibly couldn’t and that these colours did not exist in its database. I do think it could have potential to be quite dangerous tbh. How I don’t know. We are talking about paint colours here. But it was unsettling to be lied to by a computer covering up for their programming lies.

"It's frustrating to see that AI tools like ChatGPT would have read and analyzed the content multiple times."

They really don't :) That's not remotely how large language models work. This failure of understanding is all too common and slightly worrying.

I use Chatgpt a lot, but I don't like it. It does lie, and it doubles down and keeps lying once it's started. You cannot trust it for critical jobs.

But what I don't like about chatgpt is that it rewrites text or produces text unprompted. It's one thing to tell young people not to cheat by using chatgpt, but that assumes that they are going to chatgpt and commanding it to write an assignment for them. But even engaging chatgpt on a subject, or giving it some text to check for tone or grammar, will often lead to it writing something for you in the guise of being helpful.

I just cannot imagine how hard it must be to resist using chatgpt's revision of your work when it gives it to you unasked for. I'm confident enough in my own writing to recognise AI's revision is inferior, but it's unlikely students are, and I think that's going to contribute massively to plagiarism.

Chatgpt's writing is mediocre and recognisable as such to anyone who has a strong relationship with words. Lots of people don't, and I'm sure they do find it very helpful. But it's going to be hard for our young to get to the point of confident language use if chatgpt keeps insinuating their writing needs revision.

Theres a reason that a number of creators of AI are now resigned from their jobs and instead warning of the dangers of it. I hope all these people that are happily putting their feet up whilst AI writes their emails and business plans don’t have children because your children will have to fight for the few remaining jobs not done by AI; plumbers, nurses and all these admin jobs will be a thing of the past.

It is no good at all for my job (in the legal field). It completely hallucinates case law that doesn't exist, and often gets the wrong end of the stick or leaves out essential factors in referencing the ones that do exist. With long judgements it barely tries to make any real sense of them.

I say “write my some code that does <x, y, z>

and it spits out code that often just doesn’t work, but the structure is there so I have to go through and correct it, or sometimes it’s total garbage. Generally it’s code is very bulky- it’s bad at writing functions that get repeatedly called, it’ll write the same step over and over.

things like that. Mostly it just doesn’t work. Unless it’s really old- like it’s brilliant at writing excel VBA- too bad no one uses it anymore 😭

it’s also really good at telling you what a snippet of code does which can help someone hasn’t left comments about what they were doing at a certain location.

Agreed. Employers are spending a fortune on AI, not so OP can go suntan but to enhance efficiency and profits. If OP’s efficiency doesn’t increase, ie she takes on more work because she now has some tasks done by AI so she has more capacity, then they have the same output with higher input costs. So they reduce headcount and have one person covering what were 2 jobs. Simple economics.

This

We have a segregated and protected AI tool and the dataset and training code is contained within the necessary security wrapper.

I used it for a particular task to build some templates, but my final outputs are nothing like what the AI produced from my original input. That's to be expected, I'd never submit anything without heavily curating it, the initial AI proposed is far too ropey! All it did was give me a few initial ideas ad inspired me to get started on the task,

AI reminds me very much of social media, there has never been an instruction manual explaining good practice usage so the vast majority of people completely misconstrue the way to get the most out of the technology and it's like waka-mole trying to explain

I agree with posters who say you should be wary about keeping your gained time and it won’t be used elsewhere or even used to make employees redundant.

I teach and for the last two years we have used AI to write reports (as a high school teacher this saved so much time!). However, the Head didn’t like the robotic style of tone (maybe parents complained) so instead of one parents evening and a report per class, next academic year we are having two parents evenings per class. For me (teacher with no responsibility so full teaching load) that will mean conducting 14 parents evenings. What with concerts, open evenings etc I think I’m going to feel I live at school next year!