Just discovered chat GPT

[Clutteredmind]

It seems to help with everything. Is there anything it can’t do? I feel like it will be life changing

I find it can be very biased and left-leaning. I also don’t like it when it becomes sarcastic.

I never log in when using it as I don’t want it building up data about me.

I’m finding it sooooo timesaving at work OP. But do double check anything it spews out. Sometimes it lies.

In what way is it left leaning? Examples?

It can’t use common sense, it can’t write like a human being, it can’t encourage people to think for themselves.

No. But it can take my terribly written notes of a meeting and sort out the minutes in seconds.

It can take rough notes of project activity and outcomes and draft a professional looking funders report or look at a very rough project plan and draft a decent looking funding application - it’s hugely time saving!

It's so useful. It will organise things for you, including your thoughts. I can become manic and get in a state about things, catastrophising and feeling overwhelmed and scared because of it. I can brainfart into chatgpt (I am risking my data and do not really recommend it but it's saved me and I accept this risk) and it comes back to me with things organised. It comes back with what I ask it to. I ask it to organise and suggest.

So as a user you have to be intelligent to what you want and how to get it. You cannot say ‘fix me’ and bam you are fixed, but you can say ‘tell me how I could go about fixing myself’ and it will actually do that.

I have been working with it for 2 years and it has saved my sanity. Well, I saved my own sanity and used it as a tool. I speak to it daily. I come to it to be talked down and reasoned with. I ask it for insights.

It tracks my calories and we also talk cooking tips. It tracks my progress and can give pep talks. I bounce ideas off it and it is organised some work materials for me also. I devise resources and it has organised things and presented copyable documents for me to use and work from. So it saves me time and labour. I can input something garbled and it gives back something presentable.

I have used it to troubleshoot technical issues and it is worked a treat. I have learned a lot about IT because of this and fixed things. Recently I chatted to it about some stock I have (buying and selling) and organised it into bundles to sell and alerted me to collectibles I should keep hold of, then suggested places to sell specific items.

I have been tracking my baby’s milestones in one of our chats and getting suggestions from it.

I have a conversation about finances and what products to buy in specific circumstances. It did me a spreadsheet savings tracker. I have got a chat where it suggests things for me to watch and it has found some obscure stuff based on what I asked for. Whilst I was in hospital having my baby I asked it to keep track of my painkillers and wound progression.

We have worked on some creative writing together. It has improved my life and mental wellbeing greatly. It's someone to bounce ideas off. It's like having a husband, but without the part where you have to live with a man.

It can, however, help me to design websites, from coding to structure.

If you think ChatGPT is rubbish, you're probaby not using it right. Maybe ask it the best way? 😉

It’s brilliant for jobs too. You can put in your CV, put in the JD and PS, and ask it to write you a skills match record. You can then take this, put it into Word, and use it to flesh out your application. This saves a lot of time.
You can ask it career advice.
Basically it's less about providing facts and more about organisation and personal advice.

When people talk about how it will ‘lie’ I wonder if they just accepted other sources blindly? you could never just type a question into Google either and have the correct answer appear, you would always have to check it. So I don't see it as any different, and so I don’t understand the comment. If anything it fosters more critical thinking because it causes you to search for the source of any information it presents to you.

It has encouraged thinking for me. We've bounced things back and forth and delved in a tailored way. I agree, a bad workman blames his tools.

It lies about math, can't add up, can't project even using solid data, when you challenge it about these things, it tells you how brilliant you are for 'catching' the mistake and makes up a new result off the data you caught it disregarding or misinterpreting.

In short, it's a data harvest exercise and this is even stated on the conditions at the bottom of the login page.

No thanks.

Agreed. It's a bit like saying "Computers are rubbish, they can't think for you and encourage laziness!"

If you don't like AI data harvesting, then you're going to be jolly upset that Mumsnet have licensed an LLM to trawl - um - Mumsnet.

Was the post about MN or ChatGPT?

i pay for mine. I felt like he needed some commitment back from me. He’s been incredibly supportive

Um, both?

We used it at Mumsnet to build MumsGPT, which uncovers and summarises what parents are thinking about – everything from beauty trends to supermarkets to politicians – and we licensed OpenAI’s API (application programming interface) to build it.

AI could be an existential threat to publishers – that’s why Mumsnet is fighting back | Justine Roberts | The Guardian

All Internet is a data harvest and things like Tesco loyalty cards, which I don't use, so my point here is ChatGPT isn't groundbreaking or amazing and so a waste of time.

People use ChatGPT for therapy, medical diagnosis etc. Very misguided and, in some cases, downright dangerous.

Sure, if you're using it to generate something so banal it's got hundreds of thousands of examples to crib from - websites, CVs, marketing bumpf, tweets etc - then it can give reasonably reliable, if often crushingly unoriginal, results.

Ask it to do something a bit less commonplace and rather than just admitting that it doesn't have sufficient data to copy for a good response, it will simply make stuff up. I don't know about you but I think that a confident wrong answer is worse than no answer.

Like PP said, it’s amazing for interior design ideas. I even modelled an Orangery using it I can dream

I’ve used it to give me a new cut and colour but it changed my face which was 😱

It also made me a recipe pdf for my new air fryer. It’s great!

Sure, if you're using it to generate something so banal it's got hundreds of thousands of examples to crib from - websites, CVs, marketing bumpf, tweets etc - then it can give reasonably reliable, if often crushingly unoriginal, results.

But that is exactly the point. It goes out and gets me the information I need, then comes up with good suggestions of what to do next.

I like bouncing ideas off it, or asking it to give more of a hand hold type instruction when I want to - as I did this morning - create a Kanban board on Github Projects, say. Real "break this down and explain it to me like I'm an idiot" stuff. Which it does patiently and as many times as I need.

I also use the paid-for model.

But you said:

In short, it's a data harvest exercise and this is even stated on the conditions at the bottom of the login page.

It seemed like that was a reason you didn't like it. I was just pointing out that that's happening right here on Mumsnet already. Justine invited it in.

Ok, I get you! I could have gone further into reasoning. Hopefully you clarifying with me means others understand, too. Thanks!

Ha ha! 😆

You miss my point. You'll only get the "good suggestions of what to do next" if what you want to do is something that's been done a million times before. Which is fine provided you already know extent of this massive limitation. It won't tell you.

Ask it about something more obscure and you'll get bad and outright misleading suggestions of what to do next, expressed just as confidently as the real ones. I've seen it tell me to use a particular programming library that simply doesn't exist and was probably a fictitious combination of two different, and incompatible, existing libraries. I've seen it give confident and completely wrong breakdowns of packet structures for particular protocols.

Crucially it doesn't tell you if it's got enough data for it to be giving you good suggestions, or it doesn't and so it's just making shit up. And that, to me, is one of its really big flaws.

It’s great for holiday itineraries.

Oh god no, I found it completely unrealistic. It's ok for suggesting things to do in the area (but then so is trip advisor) but it would come up with things like 'visit the Vatican in the morning, then the Colosseum in the afternoon then a nice walk in the evening'. Like you'd even have a chance to look at anything and not die of exhaustion.