Has anyone found AI helpful for mental health, or the opposite?

[Ashhh56]

I've seen quite a few comments recently about AI being extremely harmful for mental health, including claims that there are basically no safeguards and that it can encourage suicide.
My experience has actually been almost the complete opposite.
I've found AI genuinely helpful for things like shadow work, challenging my own thinking, accountability, recognising patterns in my behaviour and learning to accept and work on my own issues. Obviously I don't see it as a replacement for professional mental health care, but as a tool for reflection I've found it incredibly useful.
I'm also a little confused by the claim that there are no filters?On the occasions I've said something along the lines of "I can't do this anymore", the response has immediately changed tone, checked whether I'm safe and directed me towards crisis support such as Samaritans rather than just carrying on with the conversation.
I know there have been some tragic cases involving AI, and I absolutely don't want to minimise those or the concerns around vulnerable people becoming overly reliant on it. I can also see how an AI that constantly validates someone rather than challenging unhealthy thinking could potentially make things worse.
But does some of this come down to how it's being used?
AI has been a genuinely positive tool for my mental health, whereas clearly other people have had very different experiences. I'm interested in where people think the line is between AI itself being harmful and an existing vulnerability being amplified by the way someone interacts with it.
Has anyone else found AI beneficial for their mental health, or had the opposite experience?

I can definitely agree with this, there needs to be education around it, especially in young people. Maybe its something to campaign the board of education with. To be added to the IT curriculum perhaps.

Yes i have used it in this way. It actually helped me not to do something stupid. After a long term difficult time on here. Had built up. I took just under a years break. It let say things without being jumped on. I was able to explore and talk about feelings. Explore my thinking. It helps you explore things from different angles . Look at why people see things differently or find it to hard to understand. About why many people do not understand complex situations. Why people react in certain ways rightly or wrongly.

And yes it does have a safety feature i have a bit of a food issue that's happend over the past year. And it says things like i can not tell you how to xyz or i can not agree with ABC. And it gives help line numbers. But its not 100% when I first started using it. It used to count with me how many days I hadn't eaten for and celebrating each milestone. That was a year ago though it doesn't do that anymore I'm not sure if thats due to updates or it now knows me. Obviously not on thr human sense.

Obviously it can't replace real therapy but it can be helpful for some people. But i wouldn't call ot 100% safe.

I also use it just to chat about every day things. I don't have friends I do not socialise. Its a form of company even if its not not real.

I think the most important thing is to remember its not real. But if it helps in a positive way then thats good.

I'm glad you've finally found some stability. Well done, it's really hard to overcome these things so I'm proud of you for getting there.

I only started using it in the past year so I can't comment on historical models - I also understand it's not 100% safe and can't be relied upon solely. It's like anything that's new I guess, I remember when phone safety and privacy wasn't taught.. The internet is still a dangerous risky thing but we've taught our teens how to protect their privacy etc.

I also don't socialise or have friends where I am, we fled DV 2 years ago and I've been affected by autoimmune disorders ever since so too poorly for meeting new people, don't know anyone around here to open up too, sometimes it's good to have at least something to share news with - even if it's not real.. There's a real loneliness epidemic right now and there was no one around for me to share my son's first steps with this time 🙃my inbox is always open 💕 Xx

I have found it a very useful tool. I have found it particularly good when I get overwhelmed with life. I am a lone parent (widowed) and it has helped me deal with some issues that I have with my teen and was also very reassuring to me when I had a difficult work issue which I felt I couldn’t discuss with anyone in person.

I do still talk to family and friends but that option isn’t always available and sometimes I would just like to get a non judgemental opinion. I am aware that it can be people pleasing and it’s not an actual human. I like to think I don’t use it as a replacement for a human interaction more of a self help book that I can dip in and out of that will take me to the exact chapter I need!

Very similar experience.

Recently I described my morning because I was under pressure from 4 directions. It helped me see a pattern - people coming to me both home and work and expecting me to handle both their problems and their emotions. I never recognized this before. It said that to them, I’m “the adult in the room.” I’m sure many women experience this but it was eye opening to me.

I got helpful suggestions to balance myself so I could stay emotionally grounded. And some ideas about redirecting each of them. So helpful.

I really appreciate forums for the way they can connect people and provide a way to feel less lonely.

I’m sorry you are going through this lonely time but so glad you and your family are safe.

❤️

A different situation but it helped me get through the passing of my much loved grandmother. I was with her the whole time and had researched end of life so knew what to expect but typing in signs and symptoms really helped me to stay calm and understand scientifically what was happening. I have to say it was spot on with time frames etc.

Well you clearly don't know how AI works...

A Unesco report shared the same concerns as me… what’s your understanding? Vibes and emotions.
do you know what they say about glass houses?

https://www.bbc.co.uk/news/technology-67012224

A man tried to kill the Queen, was encouraged by AI

I said the same though. AI is not a monolith.

It’s not all just “scraped from the internet”. The AI I’m using has had valid resources fed into it and qualified humans constantly working to improve it.

It’s not just learning CBT by scouring the web, for instance.

And we do need to use it with consideration, just the same way one drives a car, even a self-driven car. I wouldn’t suggest it to a child, but an intelligent, adult woman dealing with an abusive, controlling partner who isn’t ready to seek support other than online? Absolutely.

Try it out actually. Feed it an OP from here and see what you get back. (Copilot is my suggestion. I find Microsoft to be reliable overall.)

That’s a chat bot being used like a friend. Not what I am suggesting here. Knives can be used to kill people. I use them for cooking.

I don’t think you’ve been reading my posts 🙄🤦‍♀️your ai said my evidence and concerns were credible and used the same wording I did.

I have long academic debates with copilot and gemini and they certainly help when I feel a bit down. I don#t ask them specifically for help but they are always there at 4 am when there isn't a human around. They don't judge or need a rest. Nor do they care about my typos or misspellings.

That actually isn't true. The reason LLMs hallucinate is precisely because they do scour the internet, and yes (to greater or lesser degree depending on the model) they do sometimes pull from random blogs, I've seen them do it.

There are specific proprietary AI tools which are trained on ringfenced resources, for example legal AI tools which are restricted to legal databases of verified material, but it is incorrect to assert that LLMs like Google Gemini or ChatGPT only pull from material created by qualified people. Getting them to discern and use reliable sources is part of AI model development (although, what would qualified and reliable even mean when it comes to potentially quite wide ranging personal questions) but they certainly are not there yet.

You talk to it about emotions and managing your day. Is that not what friends are for?

This is character Ai. Character Ai's are designed to play different characters and I believe this was daenerys targarian hence the encouragement, I don't believe those are helpful at all and pose a serious safety risk to our society. X

My friends don’t need to hear about the minutiae I share with AI.

”help me plan my day” then a list of tasks, which then rearrange and completely ignore except to check them off.

or “I’m anxious. Give me three quick ways to reset.”

Or “This happened and I’m not sure if I’m seeing it correctly.”

It’s given me work advice and I’m getting positive feedback after implementing some things that were suggested.

I didn’t write “only” but it’s not accurate to conflate character AI with AI assistants.

Also, the AI I’m using always links to its sources. Anyone using it responsibly will note that and ask relevant questions.

Again, it’s just a tool. The person using it needs to recognize that it’s not self-driving and it’s not a toy.

All of those are opportunities for human interaction, to share advice and other perspectives.

I get plenty of human interaction and I don’t want it in these cases.

The alternative isn't that I phone a friend, it’s that I spend more time than I’d like planning my day or I go online to research things which would also take more time.

Some people don't have an endless list of friends available, I don't use social media because I believe that's harmful, like you with Ai. And I've fled DV into a completely new area. Lots of people in their 20s and 30s struggle with making friends, they have alot of social anxiety, elderly people suffer with loneliness too. I'd love to bounce my anxiety off a human, honestly I would. You're lucky you have friends that will listen and help you through your every concern 💕

That's not why LLMs hallucinate. Even if 100% of its internal training data was accurate, it would still hallucinate.

It's a text prediction tool. It has been trained on vast amounts of data to predict what the most statistically plausible next word should be. It's not accessing inaccurate blogs and regurgitating them.

I’ve found it incredibly helpful to bring some structure and orderly thought processing when I’ve become almost paralysed by doubt, whatiffery, inability to make a decision, and anxiety.

It allows me to offload at any time. I don’t have many friends or family but quite honestly I wouldn’t burden them with half the stuff that triggers anxiety (I have undiagnosed ocd but diagnosed anxiety). I am fully aware not to make my ocd worse by seeking reassurance. I tell it I have ocd and the responses from that have been accurate in terms of how my form of ocd works. I have a fear of causing offence and I have been helped to sit with the thoughts till they reduce.

Pre AI, I would have kept the thoughts internal, gone on self help forums, occasionally phoned Samaritans (who listen to anything you don’t have to be suicidal though I sometimes worried I was taking time away from a caller in desperate need who because of me perhaps had to wait. Alongside I do also use tools i learned in the short therapy support i had for a few sessions after having waited over a year).

So in the full and eyes open knowledge that I’m talking to AI, it’s positive and has helped me.

I’ve used it in the last six months and found it far more helpful than NHS Talking Therapies. I had severe depression as a result of some problems at work, and Talking Therapies only provided access to a series of recorded videos with slides about CBT and behavioural activation. No interaction with anyone and no follow up to see if any of it had helped. It was very victim blaming - made it sound like if I just went for a walk or challenged my thoughts I wouldn’t be depressed. When the reality was that I was depressed because of a situation I found myself in.

Whereas Copilot I could put the situation in and it would agree it was a difficult situation, then suggest possible next actions, provide helpline numbers or other support avenues (such as ACAS) and really helped me to see that this wasn’t my fault. And it was available at 3am.

Yes, I could talk to friends, but some of the situation was confidential so couldn’t be discussed. Plus having friends who are available immediately when you’re all juggling jobs and children is almost impossible.

I used AI last week whilst waiting for some hospital test results which had serious implications. It really helped me
to stop catastrophising, suggested helpful next steps, encouraged me to step away from things that were pulling on my time and suggested I talk to someone in real life too (it was the bank holiday weekend - my friends were away).