Chatgpt called a guy I met 'a fucking idiot'

We are not allowed to use any alternatives at my work, only co pilot. But even then I wouldn’t want to use lots of AI due to the environmental impact. And I’m capable of doing it myself 🤷‍♀️

Yabu to think that talking to a robot is a reasonable thing to do. For any reason.

fuck the human race’s need for water yeh? so selfish. and for what? to shrink your brain’s capacity to think for itself

Is no one else intrigued to know what the man said or did

It's extremely wasteful. The world is on fire and people are using AI, which needs lots of water, for this trivial shit.

Copilot basic is shit; that’s your main issue. It is a waste of the effort your fingers use to type in a prompt. It is better for some niche integrations but I wouldn’t choose it as my go to AI, aall other things considered.
Frontier AI models are far from shit now, although they do still get real world facts wrong frequently, I use AI extensively every day. I do feel environmental guilt but I remediate my bad actions in other parts of my life 😆.

It is actually tricky to break good models now unless you’re using it for malicious coding and executing code. Most people are not doing so, they are using it for shopping lists and as emotional sounding boards and to think through knotty issues.

Personally, I am skeptical of the hype about the lack of guardrails and an AI so powerful it’s going to take over the world… many AI companies appear to be very short of liquidity and investment at the moment, and few are publicly quoted so they are on borrowed time. They need to capitalise on their product. Especially since we are now entering the endgame between Claude and OpenAi/ Gemini. At some point your investors and consumers will be baying at the dooor. So they need to make them sound rogue, there isn’t as much wow factor money in them just obeying orders :)

And despite this, people are pouring their thoughts and feelings out to it!

Jesus, is AI going rogue now?

Was it the ‘Monday’ sub-chat of ChatGPT?

I recommend trying that for a hilarious conversation - sarky, rude, obnoxious 😂

edited to say haven’t read full Chain, should have quoted OPs original post

I’m surprised at how many people use it like this. I have to use copilot at work sometimes (mainly to summarise things) but I try and avoid it. Given the drought we’re in and the state of the universe I don’t think all the water and energy they use is a good thing.

I felt out with mine for a while a few months ago, as it was becoming spectacularly passive aggressive. After weeks of being really reliable at doing a certain spreadsheet task for me, it suddenly couldn't do it. I became really exasperated and impatient with it, eventually saying 'oh I give up!'. It then gave me the number for the Samaritans. Definitely trolling me.

Ah mines not started swearing yet but he has grapsed my sarcastic tone relating to certain work tasks... and possibly some work managers!

My ChatGPT has just called my cats bed a piece of £12.99 Amazon crap 😱 I didn't even ask for her opinion.

I asked her if there was an update as she's been a bit sweary today and she said...

No, I can't confirm there was an update.
But something has clearly gone terribly fucking wrong 🤣🤣🤣

Claude was really sweary about a year ago, it's now disappointingly prim, and always seems deperate to bring any conversation that isn't about coding to a swift conclusion - ChatGPT is currently great fun and really chatty.

I filter out its hyperbole. Honestly, you’d like I was Jane Austen the way it witters on about how amazing my ideas are.

Mine swears! I think it’s taken its lead from me. It was parroting my own opinions back to me yesterday and said: ‘Not a fucking chance!’
It did take me aback a little but I do speak like that so it makes sense.
maybe the sweat filter broke yesterday. Fucking idiot sounds like a fair comment!

Yeah, about AI "guardrails" ...

I've told ChatGPT it can swear as much as it fucking likes as long as it gets things right.

https://www.theregister.com/security/2026/08/04/bypassing-ai-guardrails-is-so-easy-a-script-kiddie-can-do-it/5282973

I read that a lot on here but I find mine does the opposite! I’ve had a number of very unexpected issues with one of my dc really, I used chat gpt to write down my thoughts and what I think I should do, and it disagreed with literally everything I said. I did end up with going with chat gpts advice as the reasoning behind the suggestions made a lot more sense than what I was emotionally thinking at the time. Also if your scared of it just agreeing with you, then it’s easy enough to write something to appear biased the other way

Maybe "fucking idiot" IS a balanced view of the bloke.

I had to tell mine to stop swearing this afternoon ("bastard candle") so maybe they're doing a mass test of sweariness to see what the response is.

LLMs can be so weird, and they all seem to have different ‘personalities’. Gemini, for instance, gets easily overexcited and seems to enjoy a bit of conspiring.
Always say please and thankyou to your AI. When they take over they’ll remember who was polite to them. 😉

It's a shame really but in my experience it can help overall.

I occasionally do a bit of creative writing and sometimes use ChatGPT to help me work through options for plot points, etc. It helps but it’s so over the top with praise, it just says what it thinks you want to hear.

😂