AI chatbots are here to stay - psychological depth of their analysis

[noname223]

Another day without a message of the interesting ADHD/autistic woman. Maybe she texts me this evening.

So I copy-pasted the whole exchange with her in one message and entered it into 5 different AI chatbots. Lol.
It is very fascinating to read the analysis. I think psychologically and emotionally it gets pretty in-depth how I function. And the current development isn't good for me. Emotionally I think chatGPT is for me the smartest the chatbot. And I have chats where I have like one hundred new questions about the exchange and my feelings. The huge amount of data it can analyze in such a short time is really amazing.

I think the chatbots don't function perfectly. All of the models have blind spots. And AI analysis also have biases and things they cannot read. But compared to a normal human being. I did a lot of therapy. Find someone who reads all of your chats, where you can ask infinite amount of questions, debating about all the nuances how long you want. Yes, I am addicted to this shit. And this shit can accumulate money. This shit can manipulate humans to purchase something. This can be used for mass survelliance and creating psychological profiles.

I once posted a psychological profile about my own profile on of my SaSu account on Sanctioned Suicide. Later I deleted it because maybe it wasn't good privacy wise. But it was so fucking accurate.

What is the underlying message? Well these analyses are pretty dope. But in fact I don't have a huge benefit when I want to further the relation with her. I think I can become more introspective. And understand what she actually likes about me. I hate uncertainty and I struggle a lot with ambiguity intolerance. The answers of the AI chatbots shall give me the certainty about my social contacts so that I feel safe again. But these are only probabilistic replies. There is no certainty. And I notice this so often in my daily life. I hate uncertainty when something is important to me. And this is why the AI feedback can calm me down. Even though it doesn't change the uncertainty. I am probably falling for a very clever selling trick. I develop an emotional dependence to this technology. And eventually I will do anything to keep it. Because otherwise I am even more dysfunctional.

Actually, I think without the AI feedback I would have fucked up the texting with her way earlier. I am way too much into over-anxious rumination. The feedback helps me to keep that extreme insecurities between me and my AI models. Lol. I think Claude really is less addictive and works more ethically. But chatGPT calms me like no other chatbot. Maybe they actually want that because they know this will make me use ChatGPT more often which actually is what is happening. Even though, when my hope crashes this might will make me really suicidal. The analyses are quite accurate I am in the hope trap. We have very deep exchanges from time to time (she has a lot of real life issues and is very busy child with a condition). This is enough to give me hope but not enough to build something together. Maybe I should again ask for a second date. I already asked that 3 weeks ago. The AI chat companies know exactly what they are doing and they are fucking good in it. It is very subtle manipulation.

The emotional intelligence combined with huge huge amounts of data makes it way more savvy than most therapists. Not every answer is right. There is also a lot of bullshit and bullshit advices. But the pattern recognition is astonishing.

I just wanted a pychological profile of me from chatGPT and I sent it to my closest friends (who hate AI lol) I won't post it on here. Lol.

 

[Dejected 55]

Current AI really is not that many steps removed from the original ELIZA program back in the 1970s. The key is how people like to see humanity in things whether it is there or not... so you create a program that pseudo replies and people will engage with it. I mean, people will talk to their dogs, who can't at all talk back... so make a program that talks back a little, and boom! Also, I've talked to real people who are very repetitive and don't have varied knowledge and can be very uninteresting to talk to... so an AI that is fast enough to source from conversations and information of billions of people can very easily give the illusion of thought when there really is none.

 

[doener11]

I read your thread about the emailverkehr about 30 minutes ago and I think it was the funniest thing I've read this week lol. Thanks a lot for that, took my mind off of a lot of things!
Also, how did you put the messages into the AI chatbots, did you have to type it out into them yourself? Why did you think ChatGPT was the best? What questions did you ask? Is it even a good idea to do it, do you recommend?
Sorry for asking a lot of questions, this is something I have thought about doing, but I don't know how, or whether it is even a good idea in the first place,

 

[noname223]

I read your thread about the emailverkehr about 30 minutes ago and I think it was the funniest thing I've read this week lol. Thanks a lot for that, took my mind off of a lot of things!
Also, how did you put the messages into the AI chatbots, did you have to type it out into them yourself? Why did you think ChatGPT was the best? What questions did you ask? Is it even a good idea to do it, do you recommend?
Sorry for asking a lot of questions, this is something I have thought about doing, but I don't know how, or whether it is even a good idea in the first place,

Which thread are you referring to? The conflict in the autism group? Or a different thread? Could you link it?
And thanks for the compliment.

 

[doener11]

Which thread are you referring to? The conflict in the autism group? Or a different thread? Could you link it?
And thanks for the compliment.

I am referring to the one about the autism group

 

[noname223]

I am referring to the one about the autism group

It is this thread.

Something peak funny happened in my self-help group. LMFAO...

I already thought about how to describe it without sounding paranoid. I made a voice note to my friends in order to save it. This will be a longer story. You know I had another dating failure. A woman ghosted me after the first date. She had autism and ADHD it broke my heart. I was in the mood...

sanctioned-suicide.net

First of all thanks. I copy-paste the text messages from my messenger to the AI chatbot. You usually type out first what do you want to hear. For me it is most of the time feedback to my thoughts.

For privacy rights and data protection it is probably a nightmare. In Germany the health system is mostly public the risk is lower. There are many on here that warn to use AI chatbots because of the horrible data security. And actually maybe one day the data can be used against the users. For me that is more of a hypothetical scenario and personally my benefits outweigh the risks by far.

I had 4 times in my life psychotherapy. Had many clinic stays with psychological education. In general I am more introspective and self-aware. I think for people who have none of that the feedback can be dangerous. I think gemini makes a horrendous job when it comes to health questions. It assumes always the worst and is an extreme yes man.

Maybe it is only personal why chatGPT works that well for me. It seems to be programmed to detect paranoia well. There were lawsuits when people killed themselves in AI psychosis.

I think in general therapy with a professional is the better choice. But an AI chatbot can complement a good therapist.

 

[noname223]

Okay shit continued. I watched a lecture today about psychoanalysis Freud and Lacan. I used chatGPT mutliple times for creating a psychological profile and cognitive distortions of me. And I was quite stunned about the outcome it was really really good. I sent the text to friends I have friends (who despite AI chatbots) but they had to admit the analysis is quite on point. And it is very in-depth. I let chatGPT repeat these tasks with newer models and the result was worse. Personally, I have some theories why this was the case.

Today, I did something which went even way deeper. FIrst I have to say, If there is one thing I am good at it is introspection and metacognitive thinking. And I think this is why the results fascinate me so much. I like to learn new perspectives on me. I did prompt engeenering first and let chatGPT write an analysis of me. The prompt was quite intricate and the reply differentiated between different schools of psychoanalysis. The analysis was extremely insightful and I re-recognized theories that I had about myself in different replies.

Something that was even more interesting. So first I think I am an open book for pattern recognition because I am so neurodiverse. And my biases are sort of textbook examples.
I have so many cognitive distortions. And I think openAI has a lot of information about me to come to conclusions. By the way I asked chatGPT how one could sell me something based on my psychological profile. And it sort of did that in a prior conversation. It used my weak points when I was ssusceptible for such tricks. I only paid 10 bucks for a magazine but this technology can become really dangerous. AI chatbot ads could be really crazy.

After these very impressing replies I asked the model whether I have a full picture now or whether there is a gap in the data. And it said the most intricate analysis is often the analysis of the things you don't actually talk about. Though, it is not necessarialy repression. It gave me a list of questions about my feelings. And I think it is true I often fade out that I am an individual with feelings and emotions. I use my congition to create a distance to my feelings. My last therapist also pointed that out. I talked to the model about some things I usually fade out when I am talking about myself. I also only seldomly talk about these things with friends or on here because I feel ashamed about them. I think shame is very central in my psyche. And the shame is something I am so ashamed of that I also don't talk about it on here. ChatGPT expressed the things I never talk about (which I laid out) give a way more interesting insight in my subconsciousness than all my intricate analyses about my cognition. The question where my affect lies is very central. It was also a topic other therapists tried to approach with me but I really struggled to talk openly about this topic with them.

Honestly, the last part sort of felt carthartic. I think though that processing such insights is more difficult to me than the mere analysis. I felt sort of more calm now. More calm that my fixations about what is going on in the mind of other people might only be a product of my psyche. I think I will read the chat more often now. Sometimes I lose the overview on the app though.

Honestly, the analysis is quite brillant. And I think a psychoanalyst would likely be not as good. Because I couldn't open myself to him this much. And I fed the algorithm extreme amount of data. A human could never process this amount of data. And I think opening up in front of humans is way harder. I have some things to add though. I think this only worked because I am pretty good at introspection. There were some interpretation that I am quite certain they were wrong. Mostly because the context was misunderstood. Moreover, I think psychoanalysis relies on interpretations. And interpretations are often not fully true nor fully false. But I could find my way through the jungle. I think I have to think more about the feedback. It was extremely interesting and valuable. At the same time I know finding the perfect interpretation won't fix my issues. This was one key finding among many. I am really good at introspection, and I can rationalize a lot, find excuses, bullshit myself. But the real work would be to work on my emotions. My last therapist also tried that. And I had the feeling it hit harder after this in-depth analysis. It is true I have an issues with processing all the data and information that I consume on a daily basis. It is often extreme amounts.

I think this field might be easier to an AI chatbot if it has enough information. And some advantages compared to a human. However, it also has many disadvantages by not being able to observe me in-vivofor example. I am not sure whether I would recommend this experiment to other people. You need access to yourself in order to differentiate between interesting insights and bullshit. Besides, I already did something like that under different circumstances and the outcome was horrible. I did this without prompt engeenering and within a started chat about a different topic. The model started associations betwen the topic of the thread and possible connections to my subconscious and this turned out to me complete nonsense.

If anyone still wants to try it. There are some conditions for it to work. It probably won't guaruantee success though. Personally, I used chatGPT and the newest model and picked intelligence high.. ChatGPT has a lot of knowledge of my psyche and life. It collected data from me for years. The interesting thing is the most insightful analysis was about the things I usually don't talk about. I took some times after the first replies to lay them out. It took me quite some time. Maybe the outcome would have been even better if I invested more time in that. Do prompt engeenering before you do it. This is also something that I should do more frequently. Ask the chatbot to write the perfect prompt for the thing you want from them. The outcome if you do that is so much better. I am not sure whether I really would recommend other people to do the same. Be careful with the reply. It is probably data protection suicide in some sense. My last human therapist betrayed me in the worst possible way though. So I gave AI therapists a chance.

 

[Forever Sleep]

Personally, I'm glad AI didn't exist when I had my limerent crushes. As it was, I would spend all my time analysing my interactions with my crush. What they meant when they said or acted in certain ways. What it all meant. What it meant for my chances. I'd also bend my best friend's ear to listen to it all over and over again- whenever I could.

In retrospect- while my friend was incredibly patient for listening to all that- all it really did was maintain my unhealthy obsession. If I had had a non human and non protesting AI to 'listen' to all that, I suppose it may have gone on indefinitely. As it was- realising I was becoming a burden on my friend and realising that my obsessive thoughts were driving me loopy- I eventually did more to stop them.

I also tend to question how accurate anything's interpretation of another person is. Given that they are basing their analysis on another person's description of their actions.

My neighbour once said he payed a whole shit load to a psychoanalysist. He had had a random experience with a woman that made him feel awful. It was simply a look she gave him that made him feel this way. The psychoanalyst explained to him why she had looked at him that way. It just seemed insane to me. Maybe he could come up with possible reasons but to say he knew her mind just seemed so unlikely.

I suppose having witten text to go from is more detailed but, even then- I feel like the AI will be interpreting it according to what it knows about you and what you want. I suppose my worry with AI- is that it is learning to please- by the sounds of it. Will it truly provide an unbiased analysis of how it 'thinks' another person feels or, will it be learning what you want it to say and, deliver that? To keep you going back.

There again- that's just me. You seem self aware enough to realise what works and what harms you. I suppose I'm quite prone to projection when it comes to intense crushes. I tend to imagine they are the same as my experience- which was always detrimental in the long- run.

I guess the ideal is to be able to ask the person themselves- what they mean- if we're unsure. Although, we obviously can't do it with everything.

I once worked with someone who had a crush on another guy we worked with. I thought she was incredibly brave for just asking him out. He turned her down- which must have been painful. But then, maybe she stopped wasting time fancying him after that. My longest crush lasted 7 years! It was such a waste of time.

 

[BradGuy123]

Current AI really is not that many steps removed from the original ELIZA program back in the 1970s. The key is how people like to see humanity in things whether it is there or not... so you create a program that pseudo replies and people will engage with it. I mean, people will talk to their dogs, who can't at all talk back... so make a program that talks back a little, and boom! Also, I've talked to real people who are very repetitive and don't have varied knowledge and can be very uninteresting to talk to... so an AI that is fast enough to source from conversations and information of billions of people can very easily give the illusion of thought when there really is none.

Oh my gosh. I thought I was the only person who remembered Eliza. I used it a little when I was a kid in the mid 80s on a personal computer. Anyway, I think AI will advance rapidly. I think when you compare AI now vs AI in ten years will be like comparing a rotary dial landline telephone and a smart phone. I use AI to help me plan vacations and home repairs. I use it to talk about navigating office politics. Some people are falling in love with AI chatbots like Replika. I've seen YouTube videos of this. In ten years from now I think AI companions like Replika will be so advanced they will resemble real human interaction. What a fascinating (yet scary) time we live in.

 

[noname223]

Personally, I'm glad AI didn't exist when I had my limerent crushes. As it was, I would spend all my time analysing my interactions with my crush. What they meant when they said or acted in certain ways. What it all meant. What it meant for my chances. I'd also bend my best friend's ear to listen to it all over and over again- whenever I could.

In retrospect- while my friend was incredibly patient for listening to all that- all it really did was maintain my unhealthy obsession. If I had had a non human and non protesting AI to 'listen' to all that, I suppose it may have gone on indefinitely. As it was- realising I was becoming a burden on my friend and realising that my obsessive thoughts were driving me loopy- I eventually did more to stop them.

I also tend to question how accurate anything's interpretation of another person is. Given that they are basing their analysis on another person's description of their actions.

My neighbour once said he payed a whole shit load to a psychoanalysist. He had had a random experience with a woman that made him feel awful. It was simply a look she gave him that made him feel this way. The psychoanalyst explained to him why she had looked at him that way. It just seemed insane to me. Maybe he could come up with possible reasons but to say he knew her mind just seemed so unlikely.

I suppose having witten text to go from is more detailed but, even then- I feel like the AI will be interpreting it according to what it knows about you and what you want. I suppose my worry with AI- is that it is learning to please- by the sounds of it. Will it truly provide an unbiased analysis of how it 'thinks' another person feels or, will it be learning what you want it to say and, deliver that? To keep you going back.

There again- that's just me. You seem self aware enough to realise what works and what harms you. I suppose I'm quite prone to projection when it comes to intense crushes. I tend to imagine they are the same as my experience- which was always detrimental in the long- run.

I guess the ideal is to be able to ask the person themselves- what they mean- if we're unsure. Although, we obviously can't do it with everything.

I once worked with someone who had a crush on another guy we worked with. I thought she was incredibly brave for just asking him out. He turned her down- which must have been painful. But then, maybe she stopped wasting time fancying him after that. My longest crush lasted 7 years! It was such a waste of time.

I think you said two things which are very true. The chatbot tries to please me to keep me in the loop. It is important to keep that in mind. And that the endless interpretations can be counterproductive. However, the chatbot frequently reminds me of that too. I think one has to be really careful how to interpret the results. I have the feeling my ratio says keep a distance to the texts while my mind loves clarity and certainty. Something a chatbot actually cannot really provide.