Chat GPT

I have resisted getting into the habit of using it as a search engine and outsourcing all my fact finding and decision making to it. I think those skills can deteriorate scarily easily if not used and I don't want to lose them. But I did send it a picture recently of a houseplant that wasn't doing so well and it made me a step by step plan to revive it and within a few days it was looking great again! It's also good for fixing or maintaining appliances.

Be careful. Chat GPT will just agree with what it thinks you want to hear. In that way it has guided some vulnerable people down some fatal paths. Don't outsource your mental health to a machine.

No but that's the point - on here, you're posting for advice and obviously you'll know that you'll get responses from all across the spectrum - supportive, middling, and the opposite of what you want to hear. I'm not saying that people on MN should be nasty - of course they shouldn't. But I'd rather get a spectrum of responses than have someone (in this case, ChatGPT) always agree with me no matter what.

Because ChatGPT is documented to affirm things you say even if they’re wrong.

Do you never wonder why it keeps telling you "That's a really great questions" or encouraging you all the time that you are quite right for feeling the way you do?

Do you not find that at all questionable?

I use it for scientific purposes only but I found that it actually gives sometimes radically different advice depending on what info/point of view you input and emphasize in your input. So I wouldn't trust it with relationships/psychology stuff. It just tells you what you want to hear.

there was a programme on R4 about its usefulness for therapy, and generally it was deemed that if you want a bit of an affirmative pick me up, similar to a meme, then it was OK, but if you actually wanted some therapy with a more serious issue, then it fell badly short, because, as m,ant people have already acknowledged, it tends to feed back what you wanted to hear, rather than challenge your thought processes and do any real work.

it was making people a bit narcissistic, and it reafirmed that they were always right.

https://www.bbc.co.uk/sounds/play/m002k3hk www.bbc.co.uk/sounds/play/m002k3hk]]]]
https://www.bbc.co.uk/programmes/m002dzjt www.bbc.co.uk/programmes/m002dzjt]]]]

Alternatively you could read a book, speak to a friend or just have a google - all things that could also end up giving bad information. At least AI is very upfront with the fact it can make mistakes. I use Gemini and it states that with every answer. Its warnings to be cautious and reminder that it’s not to be trusted for medical advice are even more prominent if you ask for medical advice.

AI is amazing, you just have to be fully aware of its limitations.

I get why it helps them - but that’s the thing, it’s scary how many people lean on it. Like lonely people seeking its support as they would friends, or getting mental health support instead of seeing a therapist. I’ve been there myself, kind of validating that I had been abused and I wasn’t crazy like my abuser said I was, but I question where you draw the line. It doesn’t replace actual therapy, but it is good as a tool for people who cannot access that care. I just think if so many things were accessible to us or we had decent people etc then we wouldn’t rely on AI

though , it is a daily thing for me to use. I can be putting up a picture on the wall and ask chat what the measurements are to place my picture. Or I had a colonoscopy and asked it what vegan foods I was able to eat and it tailored recipes specifically for that, including my favourite foods/cuisines. I am writing a book and asked it how to get started, pointers, help with tenses etc - I would never use it for the ideas or the aspect of writing it as that strips all creativity, but you can’t deny it’s a great educational tool to help aid hobbies. Even with writing emails, I sometimes ask it to replace a word or sentence to make it more professional. I’m torn still, because I wish I didn’t have to use it but it sure makes our lives easier

I mean, it’s obviously possible. But it was planning for me to see Cannes, Nice, Antibes and Monaco in one day 😂

My daughter and I were discussing the use of AI and ChatGPT yesterday. I used the comparison of maps - many people nowadays can't read a paper map because they are so used to using Sat Nav via their phones to get to places. While having these AI assistants might be a good thing, it's unwise to deskill an entire population.

So I would only use it for something that either doesn't really matter or that I just can't be bothered to check out using other methods (but totally could if I wanted to).

But you COULD! Using TGV to Nice, local train to Ventimiglia then Trenitalia or whatever company to Milan. Could be done in two days if you wanted!

Anyway, I wouldn't use it for timings myself. I know before I ask how much time I want in each place, (or just the one!) and ask the questions based on that.

It's not perfect I understand that, but it works for me especially regarding the local transport options and medical stuff (which I might need but hopefully not!).

I think this is helpful advice. Anything that helps us to reflect and gain insight is useful but you have to be clear with it what you want from it for example honesty, hidden perspectives, exploration etc so that you don't end up with an echo chamber, use prompts like how could I see this differently, what might I be missing, how might the other party be feeling, what might it take to change this situation, it is capable of offering you a good sounding board to a degree but as others are saying, you are in control of that so you need to 'contract' with it to agree on what sort of response it's going to give you, of you are genuinely just looking for validation and support then fair enough but it will be more robust if you ask it to adopt a neutral stance and not be afraid to gently challenge you or facilitate insight (like a therapist would do), be aware as well though, it's just a programme it will also then seek to offer you a little push back if you've asked it to, the best thing is to ask it to facilitate YOUR exploration and help you figure out YOUR decisions on what's right for you, to develop your own internal sense of what you want - otherwise you are just using a method of external validation of what's right and wrong and not developing your own inner voice and listening to yourself which is what therapy would hopefully guide you to do

It does tell you if you’re wrong. I asked it once about a dispute I was having with my partner. At the same time as validating how I was feeling, it also pointed out how some of the things I had said could have made her feel. It made me look at things from a completely different perspective.

I’ve just asked it if it uses confirmation bias. This is the answer:

Short answer: No — I don’t “use” confirmation bias.

Long answer (straight, evidence-based, no fluff):

1. What I am

1. Where bias can appear

1. What I actually do

1. What I do not do

• I don’t assume what you want to hear.
• I don’t selectively pick info that matches your belief.
• I don’t “choose” information — I follow rules about verification, safety, and evidence.

Yeah for fun I planned a trip from the uk to Paris, then Nice, Ventamiglia and then Milan and it said I could do it all in seven days 😂

I was using it for advice about getting organised in one chat and advice about a stray cat in other. I accidentally posted something about the cat in the chat about me. The confusing response of “yes the snuggle safe will be very grounding and warming for you” made me realise that ChatGPT just validates whatever you’re thinking.

I have also found this. It’s good for ideas but you do have to check separately. It is also very affirmative and so if you throw ideas and questions in, it will say it sounds fantastic.

Eek...So much wrong here!!

Cripes. You think you have undiagnosed diabetes (so presumably not on insulin) but you are having hypos?

What did ChatGPT advise?

I'm using it for running advice and so far it has been spot on. I know it's not telling me what I want to hear as I made a typo when I was telling it one of my goal paces and it immediately responded with concern as that goal pace was too fast for the evidence I had given it so far.

ChatGPT just predicts what a typical answer to your query might be based on previous answers to similar questions it has found on the internet.
It's no good for factual things either, as it just gives you what an answer looks like, so will happily invent dates, laws and studies if it needs to.

Why is it any sadder than using a blank notebook for emotional support in the form of journalling, or talking to an anonymous person on a phone line who will be working to SOP or a script of some sort, and doesn't really want to be in work talking to them on their 3rd night shift in a row when their much loved pet dog is poorly at home and the kids are hyped up and sensitive in the run up to Christmas etc etc.

I’m not sure that it’s the healthiest way of getting “another” view since it largely mirrors what you’ve put in

You’ll end up with a very biased opinion

I saw a social media post recently of a dad saying that his son had used it and encouraged him to commit suicide. It was heartbreaking