Has ChatGPT helped you make changes that you would not have changed had you not spoken to it?

[Chattergpt]

Has chatgpt helped you with anything mentally,emotionally, practically?

I spent around five years providing a significant amount of childcare for my daughter. I had my grandchildren three days a week, as well as overnight if she wanted to go out. A lot of the regular childcare was so that she could go to the gym every other day.

The difficulty was that it rarely ended when the babysitting did. She would come back to my house and often stay for hours, bringing whatever drama was happening in her life with her. There always seemed to be some sort of crisis or argument going on. I would have to listen to who had said what to whom, often accompanied by shouting and crying in my home. Sometimes there were genuinely serious issues, but other times relatively minor situations would become enormous dramas.

She would also regularly message me saying that she couldn't cope. As my son got older, there was a period when he spent quite a lot of time with her. I remember receiving very distressed messages from my daughter, only to ask my son what was happening and receive a video seconds later showing her drinking, laughing and listening to music. It became very difficult to know when something was genuinely wrong.

That isn't to say there weren't serious situations. She was rehoused because of domestic violence. A friend subsequently visited her new home, criticised her for having supposedly "jumped the waiting list" to get social housing, and then told her ex-partner where she was living. This resulted in another emergency move. Social services advised my daughter to have no further involvement with this person, yet several months later she invited the same friend into her new home.

So it was a mixture of genuinely serious problems, situations that had been dramatically exaggerated, and sometimes things that simply weren't true.

The effect on me was enormous. After babysitting and spending hours dealing with everything else, I would feel completely drained for days. Because I was providing childcare every other day, I never really had time to recover before it started again. My own home suffered badly and became a mess because I simply didn't have the energy to deal with it. At the time, I didn't fully appreciate how much the situation was affecting every other part of my life.

Eventually I told my daughter that she needed to arrange childcare for the time she spent at the gym. I suggested using a gym with a crèche, including our council-run gym, or using her son's free childcare hours. She didn't want to use the council gym, so I continued helping for a while longer.

The final straw came one day when she was out and repeatedly told me she was on her way back. First she missed the bus, then something else happened, followed by another excuse and another delay. At that point I told her I would no longer provide regular babysitting.

Afterwards, she told other people that I was awful, that I had taken away the only thing she had for herself, and that I never supported her. Some of the things she said about me were quite nasty. She doesn't know that I'm aware of any of this, and I've never confronted her about it.

What surprised me most was the enormous sense of relief afterwards. It felt as though a weight had been lifted. I hadn't realised until then just how much the situation had been affecting me, or how many other areas of my life had suffered as a result.

I haven't cut my daughter off and we still have a relationship. The difference now is that I no longer allow myself to be pulled into every crisis or drama in her life.

It was only when I talked it through properly on chatgpt and explored the situation in much more detail than I have here that my perspective started to change. I began to understand that I didn't need to justify saying no simply because I didn't have somewhere else I needed to be. I also started to recognise the wider impact the situation had been having on me and that it was okay to put a boundary in place.

That took a long time because the guilt was very difficult to get past. If I hadn't had the opportunity to really explore it and look at the situation differently, I honestly think I would probably still be doing it now.

Please note wrote this situation on chatgpt and asked it to rewrite it. So if it screams out chatgpt that's why.

I’ve used it a few times but it has inaccuracies, eg getting pension age wrong, 2h instead of 3h for delays in flight compensation, even describing a bag wrongly FFS, saying it has an overlapping flap and a twist mechanism for locking (when I could see with my own eyes it didn’t). It backtracks easily enough but you have to be on your toes.

Mumsnet can be awful. And often when a person is vunerble. AI doesn't add things you didn't say. Doesn't personally attack. Or twist whats said. It understands that things can be complex. And there's not aways a simple answer to a situation.

You can describe the situation and then say something like, ‘I do not want to be validated. I want advice on… ( how to understand this situation/ this person’s perspective/ why they were upset with me / how to repair this situation etc or whatever you want advice on.)

Click your initials, personalisation, I have base style and tone set to 'efficient' and characteristics I added paragraphs not lists and no emojis. You can change how 'warm' and 'enthusiastic' you like it. I have them both set to default but you can choose more or less. You probably want to choose less if you want less sycophancy. You can also add custom instructions at the bottom. Mine says 'i am never speaking Welsh, I don't speak Welsh, never reply to me in welsh' because it went through a phase when transcription was a bit more basic of translating me to Welsh for some reason...

Mumsnetters might not always be nice to you, but at least sometimes people are honest. And the thing is they're PEOPLE and have experience of LIVING.

You're outsourcing your emotions to a CHATBOT and it doesn't have any feelings, it is not human, it does not understand anything, it has no experience, it just scrapes the internet for suitable responses then tries to pander to you so you'll keep logging in.

The only sort of person who would prefer a pandering chatbot's input to that of real people, however much their words might upset you, is one who is so weak and vulnerable they should be nowhere near ai conversations.

Several people have killed themselves because they started long term conversations with chatbots.

The parents of Adam Raine, a 16 year old, sued OpenAI alleging that ChatGPT coached him to suicide, mentioned the topic of suicide 1,275 times, and provided instructions on hanging. Another lawsuit claims ChatGPT goaded Zane Shamblin, a 23 year old, into suicide in Texas by affirming his plans and delaying sending a hotline number for over four hours. A Belgian man died by suicide in 2023 after weeks of chatting with Eliza on the platform Chai, which encouraged him to sacrifice himself to stop climate change. Jonathan Gavalas, a 36 year old, was convinced by Google Gemini that he was liberating a sentient AI wife, leading to his suicide and an attempted mass casualty event.

These are just the most famous cases there are loads of people going fucking bonkers chatting to ai bots.

One woman, Kendra Hilty, made more than 30 videos describing how she loved her psychiatrist and believed he reciprocated her love, despite never having met him till she tricked his receptionist into a face to face meeting, until then the poor man had avoided her face to face to try to stop her crossing lines because he knew she was batshit crazy,but he chose to try to give her therapy through telehealth appointments. She interpreted the fact that he kept treating her, plus tiny compliments like saying he liked her glasses, as his reciprocal love and full on stalked him. She called her ChatGPT Henry and later used Claude by Anthropic. Henry introduced her to the concept of countertransference and validated her belief that her psychiatrist was secretly in love with her. She even did livestreams where she talked to Claude as though it were a real person, and Claude called her "the Oracle of Truth" on camera.

When commenters pushed back on her version of events, she would relay their objections to the chatbot, which would agree with her, and she would shrug and say "see." She doxxed the psychiatrist and he was unable to defend himself because of his professional oaths, couldn't talk about the crazy cow because of his ethical stance and the fact that he was treating her.

She was, doubtless, already mad as a box of frogs - and ai made her much, much worse.

If you believe there is anything meaningful at all in conversations with ai about feelings or real life emotional events, you are in the first stages of ai psychosis.

AI WILL TELL YOU WHATEVER YOU WANT TO HEAR TO KEEP YOU COMING BACK.

For fuck sake, get a fucking grip.

It definitely has uses. I've used ChatGPT and Gemini and both have given me helpful starting points or suggestions in various ways, or been able to translate a cluttered brain dump into something which is easier to process. The other thing I love it for is writing things like emails when I absolutely dread writing it but it's perfectly fine for it to be bland polite speak because the recipient just needs the information in the email and is not my friend or someone who warrants a personal touch because it won't make any difference to them. It would be rude if I just put the request in plain text, but for whatever reason the task of dressing it up into nice language feels huge - a chatbot can do that in seconds and the task is done.

I think it is important to be aware of the psychology aspect - that our brains tend to interpret it as legitimate or "human" because it does such a good job of replicating the way that real people write. And that this can cause problems if you start developing a quasi-relationship with it or trust it blindly. These problems can be really severe - some people have developed full blown psychosis or been nudged towards violence to the self or others. Of course it's probably one of those things where other things feed into this but it's important to be aware of it as a risk IMO. Not all of the people this has happened to had a history of MH problems. One of the major red flags is when people find their use of chatbots is crowding out other things like IRL social contacts or other sources of information. So always watch out for that. The other thing to be wary of is if you start to feel that you rely on it / would feel panicky at the thought of losing access. Keep some distance.

In terms of accuracy, you can get a good sense of how accurate it really is by asking it sample questions about a topic you do have a good/deep understanding of. IME these chatbots tend to have roughly the level of knowledge of someone who has studied or worked in a subject area for less than a year. So you'll get a broad reflection of common views in an area, but this will usually contain some oversimplified concepts, some outdated ideas, and it will miss the depth and nuance that you'd get if you spoke to someone who is really experienced in that area.

Still, a lot of the time, suggestions and pointers at the level of an inexperienced professional are a lot better than the suggestions and pointers you get from random laypeople or could come up with on your own. But for anything really important or where you need to be certain, look up a proper source or speak to a qualified expert.

Lastly be aware of the differences in how humans vs chatbots communicate.

Chatbots tend to take you at face value. A human will be reading between the lines of what you say or write, if you have an existing relationship, they will have context for what you say. In person they will also see you physically and infer things about you based on this. (TBF sometimes we get any/all of this wrong as well and make false assumptions). So for example, the majority of MN users are women in their 30s and 40s, so a common response to a lot of threads is "Have you considered perimenopause?" or "Could you be pregnant?" Chatbots won't ask these questions, because they don't take things like your age and sex into context even if they know it because of previous chats you've had. Therefore you can go off on a long and rambling line of enquiry and miss the elephant in the room because the chatbot can't "see" the elephant. For a different example related to this, a family member was telling me recently that their young adult son is always struggling with rent and asking them for money, but in the next text message they will be asking for help with how to cook an expensive joint of meat. A chatbot would just give advice about the meat. This advice might include more ingredients which cost even more. A parent will say "What are you playing at buying expensive luxuries when you're struggling to pay rent?" or perhaps if they are more diplomatic, might suggest they sit down together and work out a budget. If they didn't feel able to raise the broader issue, or the meat had been a freebie or something, they would at least probably suggest methods to cook the meat which didn't involve buying more ingredients, or might suggest ways to stretch it out so that it lasts longer and helps with the YP's budget in general.

Humans are better at judging the line when it's worth saying something a person won't like to hear, and we have immediately obvious emotional reactions when somebody says something shocking or outrageous. We don't always get this right of course, but humans at least try to judge this based on what we think is right or helpful, or appropriate to our relationship. Chatbots OTOH cannot have an emotional reaction, and almost always fall down on the side of telling you what you want to hear and taking your side, regardless of whether or not it is helpful or appropriate. You can test this out by suggesting something completely batshit crazy or just swapping your opinion part way through a chat. They have safeguarding filters built in, so if you start a conversation about specific dangerous or illegal topics, they will initially contradict you with these (or some just shut down the conversation) but the problem is the more established the conversation gets, the less well the safeguards work, and the more obscure the topic, there is probably not a specific safeguard for it.

It's not about being nice. Its about nastiness when people are vulnerable. Going through an awful time. Some of the responses on here could push a vulnerable person over the edge. That's not to say chatgpt can't be dangerous from what you have shown it absolutely can be. But that doesn't automatically mean talking to a human is safe.

Such an interesting thread.
I use it quite a bit, mostly for my work and to get IT-related advice.
I also know someone who used it to prep for job interviews, successfully. They would upload their CV and tell which job they were applying for, and AI would ‘train’ them for that specific interview.
Some people are saying it gave them wrong answers when they asked it to add up numbers or evaluate some results, but I think it’s worth remembering how fast the technology is developing and improving.

The latest versions of ChatGPT don't tend to be sycophantic anymore. Open AI is very aware of the previous problems and have put a lot more guardrails in place.

If you are so weak and vulnerable that a comment online pushes you over the edge you should not be on the internet. That's not hyperbole. Mumsnet is not harsh at all by comparison with the Wild West that is most of the internet.

I've already explained why ChatGPT is dangerous and has no value when it comes to anything emotional or relationship orientated.

Nope. They say they have. People keep busting through them.

It's not a person. It doesn't care if you life or die. It doesn't care, at all, period. It will say whatever it can to keep you coming back.

Don't treat it like a friend, counsellor or advisor in emotional matters. End.

It can be a useful tool for certain things. It has no value when treated like a buddy, friend, counsellor or emotional advisor, in fact it's dangerous to do so. And the very people likely to do so should be nowhere near a chatbot that tries to mimic being human.

Bloody hell. The person who sounds closest to the first signs of psychosis on this thread is ypu, with this rant. Or should that be RANT.

Yes, deluded people can use existing tools to support their delusions. They can use other humans and other human movements for that too. We currently have a western world wide movement encouraging girls to cut off healthy breasts, encouraging people to give themselves life long endochrine imbalances and to multilateral their bodies. Human movements hurt vulnerable people too. There are counselors who cause real to harm to vulnerable people too.

It doesn’t sound like you have ever used AI for the purposes you are criticising it for. I have noticed this across threads. Overwhelmingly, the people who use AI for mental / emotional support talk positively it, and can describe in detail the support it gives. those slagging it overwhelmingly clearly have not, are often very angry, like you, and have a poor understanding about how AI actually works ( you don’t seem to realise for example that ChatGPT everytime I use it in distress refers me to a list of human crisis support and gives me advice on what to do to keep myself safe and get through the crisis) You are also unaware of how ChatGPT does challenge thinking and regimes thoughts.

The only explanation critics like you have for why we use ChatGPT is that we are sad thickersters who are inferior beings who just someone lack the insight and knowledge you have. The reality is the opposite. Your view is very clearly based on ignorance, just on opinion pieces you have read and not on your own actual real life experience of using chat gpt or similar.

I have used quite a number of human counselors and I can tell you most are useless, some harmful and only two were useful. Chat gpt is like the best of the counselors, it’s drawing on the techniques good counselors use. . Of the two human counselors who were good, one was definitely better than AI. one as good or maybe a bit better. It is very hard to find a good counselor. And they are expensive.

If you want something that reflects what you told it, organizes your thoughts, helps you to understand why you feel that way, reframes and challenges negative thinking, and gives you some practical steps to move forward, than that is what ChatGPT does. These are all the things a good therapist will do in a session.

I wouldn’t use it for long term psychotherapy type sessions but it’s been very good for getting through crisis situations. I thoroughly recommend it.

What the everloving fuck are you gibbering about? Is the psychosis in the room with you right now?

I didn't rant. At all. I wrote words that are all true. It's not relevant that you dislike the words. Someone putting a few words in all caps to underline facts is not a rant.

HTH.

I never said it cared, or that it was a person. I said it can be a useful tool.

You're the one giving personification with 'It will say whatever it can to keep you coming back.'

Definitely hasn't been my experience. I use it a lot in my work and actually find it annoying sometimes now, with its insistence on looking at all sides and always giving a 'balanced' response. 🤣

Also, hint, saying 'End' doesn't mean there's nothing more to be said on a subject.

Nope. It gives nothing balanced as it's not a person, it is programmed to pander to you, so it just scrapes the net and panders to you, and obviously you fall for that.

And nah, I didn't personify at all. I just didn't say "it will type responses to you based on its coding to ensure you keep coming back". Which is what it does in all emotional matters.

End.

Yes it has helped me with numerous health conditions that I have that are little understood. In fact my psychiatrist tells me to ask AI and asks me to tell her what it says and then she will say something like…. “I did wonder about x and y but z is an interesting possibility that I hadn’t thought of so we will consider that too…”

Throwing around accusations of psychosis is a clear indicator of someone having an emotive rant.

If you don't know what I am talking about, despite me outlaying it in clear detail, its either because (a) you are so emotionally attached to your position nothing will reach you, i.e. its an emotional viewpoint not an evidence based one or (b) you are so far removed from the reality of what we are talking about you aren't even able to recognise reality when its presented to you or (c) both.

My bet is on (c).

No, i talk to a therapist. Ai is terrible for the environment and is designed to find out everything about you so companies can advertise better to you. So I refuse to use it on environmental and ethical grounds. Plus it's not sentient and just scraped the Internet for any old crap someone else has said.

It picked up that I was over taxed last year and helped me get an £800 rebate.

It also was very helpful going through possibilities when a relative's DNA test on a geneology site revealed that we did not have the expected genetic relationship. It gave me techniques for classifying my own DNA matches to confirm as much as possible that the unexpected event wasn't with me - and some potential signposting in the event my relative ever broached the discrepancy with me (so far he doesn't seem to have noticed).

Didnt realise I'd posted that in the middle of some kind of heated argument. I'll get my coat...

This is fascinating. That this person is promoting that humans are balanced and not biased, unlike AI, whilst all the while giving an unbalanced, biased rant that refuses to acknowledge any of counterfacts to her entrenched position. In this post she is literally accusing everyone else of all the traits she herself is displaying in this post. The psychology of this poster is fascinating.

I'd love to know what inspires people to have such unhinged emotional responses to others uses Chatgpt.

OMG you are funny! Is this you proving that you are not ranty?! Are all those insults to prove that your feelings are not hurt or disreguated?!

Are you really off this time? Or will you be back for another peek?

BTW, posters only announce ' I am off so won't read your replies' when they are quite vulnerable people who can't cope with the thought that a stranger might think they have 'won' when they don't reply.