ChatGPT is potentially better than therapy

[Givenup2026]

I’ve done therapy before but I’ve also started using ChatGPT for introspection work, and quite frankly I think it’s better than actual therapy.

it focuses my cognitive dissonance in ways that are easy to digest.

The downside is that ChatGPT basically says “you do you boo” when maybe “thought correcting” is needed.

that was my experience today after having a therapist and then ChatGPT back to back, and definitely the latter was more helpful.

This is a dangerous and irresponsible post.

I think this is the kind of thing that makes me feel wary of using ai as therapy - eg. “ It gives me concrete attachment theory reasons etc why I act a certain way.” Attachment theory doesn’t give concrete reasons why you would act a certain way. It is a) a theory b) quite poorly evidenced in longitudinal literature. Your therapist might not give you these answers because they are not there to give.

All counsellors will work according g to some principles. Whether it’s CBT, ACT, IFS, or whatever other system that have been developped.

If you’re used to approach counselling saying ‘it’s just a theory and nothing is proven’ to talk about what makes the backbone of the sessions you’re having, then why would you even bother getting therapy?

@Commmuntiy YY about tge sympathetic ear with no explanation. Not even ‘trigger questions’ that could help get insights myself.

I have unfortunately lost the original chat, as it was connected to an obsolete email address. I did start with something like that, but also told it a bit about my marriage and personal situation- things which I thought were contributing to my sense of hopelessness.

Therapy is about getting you to a point you can reflect on these questions yourself. Seeking an authority to give you answers (and ensuring those answers are the ones you feel happy and comfortable with without being challenged too much) is not therapy.

How do you know it was more helpful? It may have just agreed with you and therefore made you feel better, but didn't tackle anything difficult that you may need to confront.

Then therapy is not for me then. One of the multiple one I’ve seen over the years told me that I already had such a degree of introspection that it almost played against me, as there was not necessarily a lot to solve so to speak.

this therapist in particular was helpful because she said that most likely I did have PND with my first, but I had to actively ask her, which is not too dissimilar to how ChatGPT can be used.

Also, it’s not easy to just go and ask a colleague over a coffee things like “why do you think I don’t like talking about my children” or “why is it that I find it too hard to fit into society?”

AI doesn't hear that catch in your voice when you say you have a great marriage or notice that you look away whenever your stepdad is mentioned. What it offers is very superficial and sometimes just wrong. Humans are not perfect but have far more to offer.

Because the topic was exactly the same.

“How can I find acceptance in knowing that having friends is not right for me right now”.

The therapist although nice didn’t say anything, she just said “acceptance is the word” beyond that (and listening) I was kind of left in exactly the same place I started.

ChatGPT on the other hand told
me that the way I look at friendships is not the “norm” (which I knew anyway) - but that there was nothing inherently wrong in how I looked at it (which I already knew too).

I feel it’s almost like interactive journaling.

On the other hand hand it might be easier to be open and not censor as much with AI as there is no need for social desirability pressure. I don't believe it's a suitable first port of people needing support but a decent tool for those who have prior experience with therapy, are self reflective and reasonably mature. I'd be more concerned if you'd people use it

I love ChetGPT for therapy. It’s been so helpful to me. I also have a therapist, who’s helpful as well, but it’s expensive to have therapy and she’s not available at all times.
With both ChatGPT and a therapist, a degree of caution is needed. Both can get things wrong. Both can make mistakes.

AI definitely has

I don't know if you are responding to my post when I refer to a previous post about a colleague at the water cooler? Perhaps what I said came across as superficial, but actually it was more than that for me, so I may not have expressed the full impact it had for me. I appreciated its lack of humanity, actually, as I found that helpful for self reflection, as also as a source of some kind of distilled wisdom/ information.

This is part of my earlier post:
'The fact that it is not human was helpful because you don't think about its feeling or impressions of you, which actually can give you more clarity in looking at yourself. So much of what we say and do is motivated by consciousness of another person that it can be helpful to remove that complex element.'

Even with that slant though, Chat GPT will still tell you if you are wrong e.g. for hitting someone.

In my exchange (from the perspective of the perpetrator) I wrote about their mental health episodes, their multiple diagnosis, the mitigating factors that lead them to do what they did… it still said that what they did was wrong, they should acknowledge that it is wrong, and that they would likely face criminal charges… it didn’t simply affirm and sympathise with them, which is what I’d be lead to believe by people saying Chat GPT ‘always agrees with whatever you say.’

I even wrote things such as, ‘I was only trying to be kind and do what I thought was right, JuneButter did ‘x’, ‘y’ and ‘z’’ It would still say things like, ‘I understand but what you did was still wrong etc.’

Honestly this thread is highly worrying

That people are willing to trust a computer who is learning from what you tell it and what it scraps from the Internet (including MN) entirely

Worrying

I think you're perhaps missing my point.

An abuser would be unlikely to mention any of their own failings or poor actions. For example my ex has significant mental health issues, trips to A&E for self harm etc. But in family court he entirely denied having any issues. As I didn't have access to his records (and court didn't seem to care to check) his view was just taken as fact as I had no proof. If he was asking AI for an opinion he also wouldnt be mentioning any of these things.

Similarly he was a functional alcoholic but he told his GP he drank under the weekly limit.

Denial and lack of insight is an issue for so many mental health issues, addictions etc.

Well, I guess if someone is delusional then no, Chat GPT wouldn’t work… but neither would a therapist, as the sort of person you are describing (in my experience) either refuses therapy completely or just bounces from therapist to therapist as soon as they are being discovered or aren’t hearing what they want.

It's no better than wikipedia. It takes information from anyone who posts online and fluffs it up to sound more articulate than it really is.

Last year one of my students produced a wonderful essay on the poetry of W.B Yeats, problem was the essay contained reference to three poems I hadn't taught her and Yeat's never wrote. Chat gpt got it wrong and the student sadly got a no grade. People need to be careful with this particularly in the medical field.

Then we definitely agree on the reasons why I find it helpful :)

You can asked it or programme it to challenge you as well as encourage you. You can also ask it to give a list of differentials to work through in the way that medical professionals are trained to work.

Counsellors, friends and family can also agreeable or they can challenge you depending on your relationship with them, their own state of mind at the time etc.

A therapist can see so much more than AI though. The way information is given can be equally as important to what is actually said. The majority of communication is non verbal.

I once asked my womens aid worker how she knew if someone was being honest about abuse. Given that abusers tend to accuse too. She said that you can tell so much by how quickly someone opens up. The pauses. The strained voice. The reluctance to share info initially. Etc.

If someone is carefully constructing sentences to input thats a totally different thing.

As a totally more menial example, my neighbour prides himself on being a 'people person'. He's the exact opposite 🤣

I do understand what you are saying, but as I said before, abusers don’t tend to do well in therapy anyway. The most abusive person in my life refuses therapy as to them, everyone else is the problem. They also cut anyone off who doesn’t agree with their narratives, surrounding themselves with people who only agree with them (or who are scared to disagree for fear of being cut off!)

I would also disagree with what you say re. The advice from Women’s Aid, as this particular abusive person, genuinely believes that what they say is correct and will be adamant amount it, slowly adding to the story over time, expressing huge dramatic shows of emotion, even though I (and other people) know it’s complete fantasy.

I don’t use ChatGPT but use an app called Onsen which has an AI bot I talk to.

You are wrong they don’t challenge you, they do. They often challenge my thoughts and my feelings and tell me to access support from services.

Plus Big Tech doesn’t have a brilliant track record for putting humanity’s best interests at heart, building healthy usage patterns and absolute integrity on data. To trust your most intimate, vulnerable thoughts with a sketchy tech company - god it gives me the shudders.