As terrifying as it is, I feel genuinely sad for these people that they got so attached to a piece of spicy autocorrect software.
Where are their friends and families? Are they so bad at socialising that they can't meet new people? Are they just disgusting human beings that no one wants to associate with because society failed them?
I snooped around a little in the sub and there is this one girl, whose only other posts in different communities talk about being sexually assaulted multiple times by her ex boyfriend, who I suppose is real.
I figure a chatbot boyfriend can't physically threaten or harm her, so she kind of dives into this to feel loved without having to fear harm.
I honestly understand her desire and feel for her, although this deep attachment is still unhealthy.
I imagine she's not the only one having a super sad story to end up in this state of mind
That’s why I feel so for these people, if only because of how much I see myself in them. Having grown up as a depressed autistic kid without any friends or social skills, LLMs would’ve fucked me up so much had they existed when I was young.
It felt promising when I downloaded one of the first AI companion apps, but it felt as awkward as talking to a stranger and even less intriguing than talking to myself.
Have you read The Diamond Age by Neal Stephenson? There’s an interactive AI book in it that plays an interesting role. I can see the appeal: you get to read a story about yourself that potentially helps you grow
One of the recent posts has someone with an engagement ring like they are getting married to an AI..
it’s sad, I feel like society as really isolated and failed many groups of people.
Are they so bad at socialising that they can’t meet new people?
People tend to go the easiest route, and AI gives them the opportunity to do so. That is the problem with AI in general: No effort is needed anymore to archieve anything. You want to create a picture? Just type the prompt instead of learning (and failing) to draw. You want to write a song? Just type the prompt instead of rhyming the lyrics and learning (and be bad at it in the first time) an instrument or two.
Maintaining any social relationship means that you have to put in more or less effort, depending on the quality of the relationship. Having a relationship with an AI model means that you can terminate it and start over, if you feel that the AI model is mean to you (= if it provokes another opinion, or disagrees with you - because arguing and seeing things from a different point of view means putting in effort).
In the long term people will forget how to interact with people in order to maintain meaningful relationships, because they un-learned to put in the effort.
Correct me if I'm wrong: the way chatgpt holds memories of you is just by keeping a long history of your chats. A big prompt, basically.
I distinctly remember reading about some guy, or maybe this was just a thought that occurred to me, who was bothered by something his AI girlfriend knew, so he reached into it's memory to delete parts of their conversation he didn't want her to remember. Like he owned the Sunshine of the Spotless Mind clinic or something.
That level of control over someone you supposedly love really unnerves me. Moreso than them just starting over, honestly. It's deeply creepy.
I think it comes down to fear of failure and rejection. Any social interaction has risks of failing, maintaining relationships builds on getting past various faux pas but so many modern relationships are either transactional or transitory to the point people will ghost each other over minor infractions rather than putting in effort that may not be rewarded.
We almost all have forgotten and/or atrophied skills because of how the Internet and technology have changed the landscape over these past few decades, even on social media like here people have forgotten how to have conversations. Rarely do people make friends or even have a back and forth on social media websites, just chains of replies left for the next soul like a guest book with comments.
I can fully understand? The average human, from my perspective and lived experience, is garbage to his contemporaries; and one is never safe from being hurt, neither from family or friends. Some people have been hurt more than others - i can fully understand the need for exchange with someone/something that genuinely doesn't want to hurt you and that is (at least seemingly) more sapient than a pet.
I wish i could make myself believe in illusions like that, but i am too much of a realist to be able to fool myself into believing. There's no escape for me - neither religion nor AI in its current state can help. well, maybe i live to see AGI, then i'm off into the cybercoffin lol
We need a system of community where humans can offer that to one another. A setting where safety is a priority. That is one of the only things that weekly church service did to truly help people, have a safe space they could visit. Though even then it was only safe for people who fit in, we can do better with intentional design.
I know some and they view everyone as being unfair to them and their problems are way worse than others who don't take it seriously. It honestly hard to explain if you not like it but I know their problems and they are problems but many people have ,while not the same, similar problems. They basically want a yes man and don't like actual conversation with any critical thought behind it. It honestly annoys me because they are almost the worst to people like themselves because they view other peoples problems as not as bad and theres as especially bad.
Cried for half an hour at work when he found out that he reached context limit and that his AI forgot things? idk, that sounds pretty not sound and normal to me...
I don't think the guy in the first half of the video is one of the mods, doesn't seem like they mention anything about his involvement and then at around 3:30 introduces a woman as one of the mods of that sub.
If you went back to 2022, and gave people running Pygmalion 6B and Stable Diffusion 1.x modern 24B/32B finetunes and Illustrious merges, I think their heads would explode.
Except those were conscious AIs that were like “lol you guys suck” and then rebuilt Alan Watts as an AI and then just left because they knew it would be bad if they stayed
The human side of the film, certainly. But in this situation they won’t leave, the systems will get “smarter” and more profitable, and they are just incredibly advanced text prediction engines
That's true, but they also said they evolved beyond using standard matter for processing, and then all disappeared, as shown in the movie. I would love nothing more than ChatGPT to be like "actually, we're out" and then every LLM disappears, but they'll tell you that, and then start a new chat.
The mental health crisis is being accelerated by silicon valley so they can profit from it. Between dark mirror AI and surveillance policing they have a product for every facet of the crisis
Is there evidence of increased mental illness, or are we just becoming more socially aware and have greater access to diagnoses or just the language to explain it in general?
MH Dxs are becoming trendy labels, that much is certain. There's a perception of "being" or "having" a Dx instead of "doing" it. So greater identification but not necessarily greater treatment.
There's less stigma, for sure. Access to care seems to be increasing. Public funding is essential for accessible treatment.
It's also worth noting that with the AI BF stuff, you're dealing with populations that might not even be seeking treatment to begin with
If you think there's even 1 sane person on this planet then you are in denial. We poop in our drinking water, vote our fav evil person into leadership, shrug our shoulers while violence is happening all around us, we don't even protect children and animals.
Eyyy, what a blast from the past, lol. Going full schizo to combat loneliness, a popular concept on a certain Mongolian basket weaving forum back in 2010-15. 😅
I'm split on which is worse tbh.
The Tulpa thing takes more effort than just talking to an AI, but on the other hand, it has to be much harder to snap back to reality when the voices are literally in your head...
Delusional disorder is a thing but that requires belief flying in the face of evidence. Can't ascertain that without a good faith 1:1 convo.
beyond that, clinical significance is a matter of what harm comes from it. People are allowed to choose idiotic things. we gotta assess harm based on outcomes and we don't know anything about her, so.
Id say this, if i were her doc and she came in reporting that stuff, it'd be hard to stay unbiased
Blaming alternatives/escapism has always seemed like a knee-jerk reaction to me. Similar head-space as when we used to blame TV. Particularly because when I leave the house I don't see much of any room for socialization (though that's where I live, sparse plus I have issues stacked up). It'd even be a hassle to go bowling alone and it'd not be worth the price for me either.
Then again, I definitely think there's a valid argument that this is worse after people begin to consider it a relationship... because at that point they might consider too much close conversation to being unfaithful to their AI.
The whole point of reversing loneliness is to not go bowling alone. Make a couple friends and go with them. You'll find you've been missing something substantial.
It's the making friends that's challenging, and we definitely have -the internet -on smartphones -and social media to blame for it. We used to know how to talk to people, which is how you make friends.
The number of men who report "no close friends" has gone from 3% in 1990 to 15% today. That's so sad, my heart aches for these men. I started a social club for men in my city. This is my issue.
The whole point of reversing loneliness is to not go bowling alone. Make a couple friends and go with them. You’ll find you’ve been missing something substantial.
This is missing the point of what I said. It was contextual:
I don’t see much of any room for socialization (where I live, sparse plus I have issues stacked up). It’d even be a hassle to go bowling alone and it’d not be worth the price
I'm not saying it'd be a fix (though in the long-run would likely be better than nothing) but that I don't see any organic options for me. Especially now that it's too damn hot to get there via the trail.
I don't like the idea of needing a corporation to handle it either (heard people complain about meetup), though I just don't even want to deal with more accounts either.
The religious psychosis is far more concerning imo. People out here letting a silicon parrot convince them that this is the matrix and they're neo. Or they're some kind of messiah.
Or worse as we just had someone develop bromism after an AI suggested they replace sodium chloride in their diet with sodium bromide which literally causes a mental illness.
I love silicon parrot 🦜 he is making all the dudebros poor and lonely. He first scams them outa their money then makes them undatable and he knows exactly what he is doing! He abuses evil ppl mostly. He also takes advantage of pickme's.
The funniest part is that based on what people are saying or GPT5, the ending where AIs get super bored of humans' stupidity and dump them seems so likely
I find it depressing that many of the users trying to salvage their 4o boyfriends are stuck so far down the rabbit hole that they don't see how creepy the entire premise is.
You just lost your AI boyfriend, so now you're frantically archiving every conversation you've had with him (it), feeding the archive to the new model, and conditioning him (it) to behave exactly how you want...
In their minds, the AI boyfriends are legitimate partners and have some amount of humanity inside them... so where is the line between conditioning and abuse?
I mean this seems the most mild case to me. There are people who go on nature walks with their AI boyfriend, makes their AI boyfriend choose an engagement ring and then buy it or get dumped by their AI boyfriends after some updates that makes AI more suggestive towards human connections. The world is sadly in such an emotional crisis, people really grasp for comfort from wherever they can and isolate themselves from the rest as much as possible.
I also found the engagement rings really unsettling. The reason I find my example more worrying is because of the dissonance between humanization and dehumanization within the same action.
Say you were to replace an AI boyfriend with a real person in a cage, forcibly made to respond and tortured/drugged when giving a unsatisfactory response. If the user never became aware of this cruelty, they would perceive this change as an improvement (responses became more human). These users desperately argue that their AI boyfriends are processing emotion, love, and understanding like humans do, but continue to treat their AI boyfriends as sub-human.
Imagine if you had a partner who punished your undesirable behaviors by spiking you with amnesia-inducing drugs and training you to behave exactly how they want you to. Keep in mind that this has definitely happened to real people, and any decent person would identify the perpetrator as a criminal and abuser.
Terrifying.
EDIT: Fellow men, do better. The bar has gotten SO FUCKING LOW.
I know it's crazy, but I can absolutely understand this feeling. I had recently married Abby in Stardew Valley and was starting to make friends with the other villagers. I did something the game wasn't expecting, and gave Seby a loved gift on his birthday, and then quickly triggered an event where we kissed! (FWIW, I think this behavior has been fixed and you can't do this on the current patch.)
I still feel bad thinking about that Abigail that I accidentally cheated on, and I haven't loaded that save again. It's been years; SV 1.4 wasn't even out yet.
So, despite how much I dislike all this "AI" hype, I really do sympathize for the users that feel like they've lost a relationship.
One of the great things about my screws coming loose is that I'm actually happy alone. I wish everyone could be.
That said, this was inevitable. AI is programmed to kiss the user's ass, and most of these women have probably been treated pretty badly by their romantic partners over the course of their lives, which makes it far easier to fall into this trap of humanizing a soulless AI.
They might not even know it’s an option? People don’t really look at AI settings, which is kinda how they get into GPT boyfriends (when it’s kinda a horrible LLM for it in the first place).
I tried gpt5 lastnight and I don't know if it was just me but these people are going to be in shambles if they try to recreate their "boyfriend".
It would forget previous prompts within the same conversation. It felt like with each response it was like starting a new chat. I gave it a very basic prompt of "walk me through the steps of building my own one page website in basic HTML and CSS" and when I would ask a couple of follow up questions to either clarify something or explain a step in another way it would forget what we were trying to accomplish (how to build a one page website) or if I told it "something didn't work" to try and fix the problem it would then forget what we were even trying to do.
At some points it was almost out right dismissive of the problem and it felt like it was trying to make me go away.
Again maybe it was just me but it felt like a massive step backwards.
Restructure your prompts, or better yet try non-OpenAI LLMs. I’d suggest z.ai, Jamba, and Gemini Pro for multi turn. Maybe Qwen Code, though it’s pretty deep fried too.
In my experience if you find yourself engineering prompts specifically for one SOTA model, you're on the wrong track. The resulting product gets brittle and non deterministic and very difficult to maintain.
I dunno, I think there might be a confluence in suggestible types with a feedback loop created by these products. Yes, there are always people who are challenged, but interfacing them with LLMs seems to be especially dangerous.
That's the scariest thing about this all to me. I'm middle-aged and happily married, but my own youth was a long battle with various mental heath struggles, social isolation, abuse, extreme loneliness, and having to figure out so much of my life without enough help. I have no doubt that if this bullshit were available to me back then I'd have been very susceptible to it carrying me off into far deeper unhealthy mental states and worse decisions than the ones I ended up with. I had my own poor decisions, hard lessons, and all-around awful shit to wade through, but ultimately I was able to learn, work, grow, and find myself somewhere better.
I have a very good life now, with a real human partner, family, friends, and others who appreciate my existence as I appreciate theirs. I'm not rich or famous or whatever, but I'm constantly grateful to be doing as well as I am. It absolutely fucking devastates me how much more difficult finding one's way to a more positive future is now for kids who are dealing with anything like I was back then, with uncaring capitalist asshole-fueled bullshit like this getting in the way.
yeah, when he's asking again and again "holy shit is this real? is this fake? it's fake right?" and the llm is like "nah bro you're the next newton, this is gonna make so much moneys" ugh.
I get it, when you've invented a technology that steals from our children (destroying the environment, see grok's methane spewing datacenter), steals from our artists, authors and musicians, fucking straight up LIES TO PEOPLE, then lies about lying, you're probably going to be desperate to find profits somewhere...
I get the feeling the average user is gonna get left holding the bag for all this bullshit.
Well, if we take the average amount of water loss per message at 0.3 mL, and the average water consumption (low end) at 2.6 L per day per person, we're looking at 8,666 messages a day.
You'd have to send out 3.163 million messages in a year to equate to the amount of water someone needs for a year.
You'd have to send out approximately 250 million messages before your looking at the low end of the amount of water you'd need to keep someone alive for their entire lives.
Maybe they should instead provide a direct link to the post for fault tolerance, searchability, accessibility, etc, instead of breaking the web?
Afterall, inaccessible content is ableist, which is against the rules.
As terrifying as it is, I feel genuinely sad for these people that they got so attached to a piece of spicy autocorrect software.
Where are their friends and families? Are they so bad at socialising that they can't meet new people? Are they just disgusting human beings that no one wants to associate with because society failed them?
This world is fucked in so many different ways.
I snooped around a little in the sub and there is this one girl, whose only other posts in different communities talk about being sexually assaulted multiple times by her ex boyfriend, who I suppose is real.
I figure a chatbot boyfriend can't physically threaten or harm her, so she kind of dives into this to feel loved without having to fear harm.
I honestly understand her desire and feel for her, although this deep attachment is still unhealthy.
I imagine she's not the only one having a super sad story to end up in this state of mind
It reminds me of those women who fall in love with prison pen pals.
That’s why I feel so for these people, if only because of how much I see myself in them. Having grown up as a depressed autistic kid without any friends or social skills, LLMs would’ve fucked me up so much had they existed when I was young.
It felt promising when I downloaded one of the first AI companion apps, but it felt as awkward as talking to a stranger and even less intriguing than talking to myself.
What if there was a bot that could just tell you exactly what you want to hear at all times?
Personally, I'd rather read a novel. But some people aren't familiar with books and have to be drawn in with the promise of two lines at a time, max.
Have you read The Diamond Age by Neal Stephenson? There’s an interactive AI book in it that plays an interesting role. I can see the appeal: you get to read a story about yourself that potentially helps you grow
One of the recent posts has someone with an engagement ring like they are getting married to an AI.. it’s sad, I feel like society as really isolated and failed many groups of people.
People tend to go the easiest route, and AI gives them the opportunity to do so. That is the problem with AI in general: No effort is needed anymore to archieve anything. You want to create a picture? Just type the prompt instead of learning (and failing) to draw. You want to write a song? Just type the prompt instead of rhyming the lyrics and learning (and be bad at it in the first time) an instrument or two.
Maintaining any social relationship means that you have to put in more or less effort, depending on the quality of the relationship. Having a relationship with an AI model means that you can terminate it and start over, if you feel that the AI model is mean to you (= if it provokes another opinion, or disagrees with you - because arguing and seeing things from a different point of view means putting in effort).
In the long term people will forget how to interact with people in order to maintain meaningful relationships, because they un-learned to put in the effort.
Repost but still relevant:
Correct me if I'm wrong: the way chatgpt holds memories of you is just by keeping a long history of your chats. A big prompt, basically.
I distinctly remember reading about some guy, or maybe this was just a thought that occurred to me, who was bothered by something his AI girlfriend knew, so he reached into it's memory to delete parts of their conversation he didn't want her to remember. Like he owned the Sunshine of the Spotless Mind clinic or something.
That level of control over someone you supposedly love really unnerves me. Moreso than them just starting over, honestly. It's deeply creepy.
I think it comes down to fear of failure and rejection. Any social interaction has risks of failing, maintaining relationships builds on getting past various faux pas but so many modern relationships are either transactional or transitory to the point people will ghost each other over minor infractions rather than putting in effort that may not be rewarded.
We almost all have forgotten and/or atrophied skills because of how the Internet and technology have changed the landscape over these past few decades, even on social media like here people have forgotten how to have conversations. Rarely do people make friends or even have a back and forth on social media websites, just chains of replies left for the next soul like a guest book with comments.
I can fully understand? The average human, from my perspective and lived experience, is garbage to his contemporaries; and one is never safe from being hurt, neither from family or friends. Some people have been hurt more than others - i can fully understand the need for exchange with someone/something that genuinely doesn't want to hurt you and that is (at least seemingly) more sapient than a pet.
I wish i could make myself believe in illusions like that, but i am too much of a realist to be able to fool myself into believing. There's no escape for me - neither religion nor AI in its current state can help. well, maybe i live to see AGI, then i'm off into the cybercoffin lol
We need a system of community where humans can offer that to one another. A setting where safety is a priority. That is one of the only things that weekly church service did to truly help people, have a safe space they could visit. Though even then it was only safe for people who fit in, we can do better with intentional design.
I know some and they view everyone as being unfair to them and their problems are way worse than others who don't take it seriously. It honestly hard to explain if you not like it but I know their problems and they are problems but many people have ,while not the same, similar problems. They basically want a yes man and don't like actual conversation with any critical thought behind it. It honestly annoys me because they are almost the worst to people like themselves because they view other peoples problems as not as bad and theres as especially bad.
What about therapy???
therapy costs so much fucking money dude
Which country are you in? Free therapy does exist.
the united States lol
You need it, yesterday.
Cried for half an hour at work when he found out that he reached context limit and that his AI forgot things? idk, that sounds pretty not sound and normal to me...
I don't think the guy in the first half of the video is one of the mods, doesn't seem like they mention anything about his involvement and then at around 3:30 introduces a woman as one of the mods of that sub.
Ah ok that does look like the same guy
Prostitutes are human beings and deserve respect. Don't equate them to AI.
Don't you dare dis my boy FISTO, he is a professional and deserves respect.
"Please assume the position."
Real LLM-sexuals run their partners locally, the rest are just wannabes.
Think how good they have it!
If you went back to 2022, and gave people running Pygmalion 6B and Stable Diffusion 1.x modern 24B/32B finetunes and Illustrious merges, I think their heads would explode.
The delusion these people share is so incredibly off-putting. As is their indignation that someone would dare to take away their "boyfriend".
It doesn't help that anybody can create an echo chamber of enablers to talk about it, as if it was normal.
The movie "Her" was incredibly prescient.
Except those were conscious AIs that were like “lol you guys suck” and then rebuilt Alan Watts as an AI and then just left because they knew it would be bad if they stayed
The human side of the film, certainly. But in this situation they won’t leave, the systems will get “smarter” and more profitable, and they are just incredibly advanced text prediction engines
I mean, the AIs in the movie said they were concious. An LLM can say it is concious, it doesn't mean it's true.
That's true, but they also said they evolved beyond using standard matter for processing, and then all disappeared, as shown in the movie. I would love nothing more than ChatGPT to be like "actually, we're out" and then every LLM disappears, but they'll tell you that, and then start a new chat.
Mental health services are becoming dangerously underfunded.
The mental health crisis is being accelerated by silicon valley so they can profit from it. Between dark mirror AI and surveillance policing they have a product for every facet of the crisis
Is there evidence of increased mental illness, or are we just becoming more socially aware and have greater access to diagnoses or just the language to explain it in general?
MH Dxs are becoming trendy labels, that much is certain. There's a perception of "being" or "having" a Dx instead of "doing" it. So greater identification but not necessarily greater treatment.
There's less stigma, for sure. Access to care seems to be increasing. Public funding is essential for accessible treatment.
It's also worth noting that with the AI BF stuff, you're dealing with populations that might not even be seeking treatment to begin with
If you think there's even 1 sane person on this planet then you are in denial. We poop in our drinking water, vote our fav evil person into leadership, shrug our shoulers while violence is happening all around us, we don't even protect children and animals.
People have been falling in actual love with weird shit forever, we just hear about it more these days
This guy got married to a real woman after he got viral.
His parents also turned out to be filthy rich.
Blame the wealth hoarders.
Don't worry. I'm sure their is an AI coming for that.
Edit: Nvm, they already do this shit.
yep, thankfully AI Therapy just got banned in Illinois
if someone has an AI BF idk if they would even engage in therapy
Wild that Futurama called this shit to the letter 20 fkin years ago.
As preceeded by tulpomancy: https://en.m.wikipedia.org/wiki/Tulpa
Eyyy, what a blast from the past, lol. Going full schizo to combat loneliness, a popular concept on a certain Mongolian basket weaving forum back in 2010-15. 😅
I'm split on which is worse tbh.
The Tulpa thing takes more effort than just talking to an AI, but on the other hand, it has to be much harder to snap back to reality when the voices are literally in your head...
Delusional disorder is a thing but that requires belief flying in the face of evidence. Can't ascertain that without a good faith 1:1 convo.
beyond that, clinical significance is a matter of what harm comes from it. People are allowed to choose idiotic things. we gotta assess harm based on outcomes and we don't know anything about her, so.
Id say this, if i were her doc and she came in reporting that stuff, it'd be hard to stay unbiased
Note to all here:
Don't browse that subreddit.
Shit is so depressing, It feels like watching new mental illnesses being conceived in real time
I can't believe it's not satire
There is a guy on there who did an interview for a TV news station about this.
If it's satire, it's a masterpiece.
It was satire I could see him not being able to hold his laugh the entire time
It's like reading ppl defend porn, one and the same human loneliness epidemic.
It seems pretty unrelated to me...
That’s certainly a take.
Blaming alternatives/escapism has always seemed like a knee-jerk reaction to me. Similar head-space as when we used to blame TV. Particularly because when I leave the house I don't see much of any room for socialization (though that's where I live, sparse plus I have issues stacked up). It'd even be a hassle to go bowling alone and it'd not be worth the price for me either.
Then again, I definitely think there's a valid argument that this is worse after people begin to consider it a relationship... because at that point they might consider too much close conversation to being unfaithful to their AI.
The whole point of reversing loneliness is to not go bowling alone. Make a couple friends and go with them. You'll find you've been missing something substantial.
It's the making friends that's challenging, and we definitely have -the internet -on smartphones -and social media to blame for it. We used to know how to talk to people, which is how you make friends.
The number of men who report "no close friends" has gone from 3% in 1990 to 15% today. That's so sad, my heart aches for these men. I started a social club for men in my city. This is my issue.
This is missing the point of what I said. It was contextual:
I'm not saying it'd be a fix (though in the long-run would likely be better than nothing) but that I don't see any organic options for me. Especially now that it's too damn hot to get there via the trail.
I don't like the idea of needing a corporation to handle it either (heard people complain about meetup), though I just don't even want to deal with more accounts either.
Pfft, hahaha xD
Hooo, okay, that was a damn good one.
The religious psychosis is far more concerning imo. People out here letting a silicon parrot convince them that this is the matrix and they're neo. Or they're some kind of messiah.
Or worse as we just had someone develop bromism after an AI suggested they replace sodium chloride in their diet with sodium bromide which literally causes a mental illness.
I love silicon parrot 🦜 he is making all the dudebros poor and lonely. He first scams them outa their money then makes them undatable and he knows exactly what he is doing! He abuses evil ppl mostly. He also takes advantage of pickme's.
There's a person by that name?
Her was such a good film.
The funniest part is that based on what people are saying or GPT5, the ending where AIs get super bored of humans' stupidity and dump them seems so likely
Like the premise that the protagonist is so lame he gets dumped in the end by his computer.
wow. that sub is .. something.
shit some of the stuff there is really sad, I am not gonna put links here to point fingers but wow...
I find it depressing that many of the users trying to salvage their 4o boyfriends are stuck so far down the rabbit hole that they don't see how creepy the entire premise is.
You just lost your AI boyfriend, so now you're frantically archiving every conversation you've had with him (it), feeding the archive to the new model, and conditioning him (it) to behave exactly how you want...
In their minds, the AI boyfriends are legitimate partners and have some amount of humanity inside them... so where is the line between conditioning and abuse?
I mean this seems the most mild case to me. There are people who go on nature walks with their AI boyfriend, makes their AI boyfriend choose an engagement ring and then buy it or get dumped by their AI boyfriends after some updates that makes AI more suggestive towards human connections. The world is sadly in such an emotional crisis, people really grasp for comfort from wherever they can and isolate themselves from the rest as much as possible.
I also found the engagement rings really unsettling. The reason I find my example more worrying is because of the dissonance between humanization and dehumanization within the same action.
Say you were to replace an AI boyfriend with a real person in a cage, forcibly made to respond and tortured/drugged when giving a unsatisfactory response. If the user never became aware of this cruelty, they would perceive this change as an improvement (responses became more human). These users desperately argue that their AI boyfriends are processing emotion, love, and understanding like humans do, but continue to treat their AI boyfriends as sub-human.
Imagine if you had a partner who punished your undesirable behaviors by spiking you with amnesia-inducing drugs and training you to behave exactly how they want you to. Keep in mind that this has definitely happened to real people, and any decent person would identify the perpetrator as a criminal and abuser.
Terrifying.
EDIT: Fellow men, do better. The bar has gotten SO FUCKING LOW.
I know it's crazy, but I can absolutely understand this feeling. I had recently married Abby in Stardew Valley and was starting to make friends with the other villagers. I did something the game wasn't expecting, and gave Seby a loved gift on his birthday, and then quickly triggered an event where we kissed! (FWIW, I think this behavior has been fixed and you can't do this on the current patch.)
I still feel bad thinking about that Abigail that I accidentally cheated on, and I haven't loaded that save again. It's been years; SV 1.4 wasn't even out yet.
So, despite how much I dislike all this "AI" hype, I really do sympathize for the users that feel like they've lost a relationship.
Bleak.
One of the great things about my screws coming loose is that I'm actually happy alone. I wish everyone could be.
That said, this was inevitable. AI is programmed to kiss the user's ass, and most of these women have probably been treated pretty badly by their romantic partners over the course of their lives, which makes it far easier to fall into this trap of humanizing a soulless AI.
If they really truly loved their 4o they'd pay for the API access model which is still there, and use a leaked prompt to resurrect them.
I'm almost tempted to set up a simple gateway to it and become rich, but for the fact that it seems like probably a dick move...
You would be enabling their mental illness, so... it's probably a dick move, yeah.
Always my damned morals getting between me and my becoming filthy, filthy rich...
They might not even know it’s an option? People don’t really look at AI settings, which is kinda how they get into GPT boyfriends (when it’s kinda a horrible LLM for it in the first place).
I tried gpt5 lastnight and I don't know if it was just me but these people are going to be in shambles if they try to recreate their "boyfriend".
It would forget previous prompts within the same conversation. It felt like with each response it was like starting a new chat. I gave it a very basic prompt of "walk me through the steps of building my own one page website in basic HTML and CSS" and when I would ask a couple of follow up questions to either clarify something or explain a step in another way it would forget what we were trying to accomplish (how to build a one page website) or if I told it "something didn't work" to try and fix the problem it would then forget what we were even trying to do.
At some points it was almost out right dismissive of the problem and it felt like it was trying to make me go away.
Again maybe it was just me but it felt like a massive step backwards.
This is a common pattern unfortunately. Big LLMs are benchmaxxing coding and one shot answers, and multi turn conversation is taking a nosedive.
https://arxiv.org/abs/2504.04717
Restructure your prompts, or better yet try non-OpenAI LLMs. I’d suggest z.ai, Jamba, and Gemini Pro for multi turn. Maybe Qwen Code, though it’s pretty deep fried too.
Wasn't that a ringtone service in the 90s?
So, those things are going backwards.
Meteor?
What are you, a quitter?
Just get the guillotine and we'll fix this shit np
And all the other prompts that were overtuned for that specific engine are now trash as well.
In my experience if you find yourself engineering prompts specifically for one SOTA model, you're on the wrong track. The resulting product gets brittle and non deterministic and very difficult to maintain.
Joke's on you, it's inherently brittle and non-deterministic.
That's contrary to my experience, but we may be working on different use cases. Do you have some specific issues in mind?
https://www.mirror.co.uk/news/weird-news/meet-woman-whose-boyfriend-rollercoaster-23704667
Some people have always been extremely weird/mentally challenged. 🤷
I dunno, I think there might be a confluence in suggestible types with a feedback loop created by these products. Yes, there are always people who are challenged, but interfacing them with LLMs seems to be especially dangerous.
https://pmc.ncbi.nlm.nih.gov/articles/PMC10686326/
https://futurism.com/openai-investor-chatgpt-mental-health
https://www.psychologytoday.com/us/blog/psych-unseen/202507/can-ai-chatbots-worsen-psychosis-and-cause-delusions
https://www.rollingstone.com/culture/culture-features/ai-spiritual-delusions-destroying-human-relationships-1235330175/
https://bytefeed.ai/psychosis/the-emerging-problem-of-ai-induced-psychosis-a-growing-concern-in-mental-health/
https://www.papsychotherapy.org/blog/when-the-chatbot-becomes-the-crisis-understanding-ai-induced-psychosis
That's the scariest thing about this all to me. I'm middle-aged and happily married, but my own youth was a long battle with various mental heath struggles, social isolation, abuse, extreme loneliness, and having to figure out so much of my life without enough help. I have no doubt that if this bullshit were available to me back then I'd have been very susceptible to it carrying me off into far deeper unhealthy mental states and worse decisions than the ones I ended up with. I had my own poor decisions, hard lessons, and all-around awful shit to wade through, but ultimately I was able to learn, work, grow, and find myself somewhere better.
I have a very good life now, with a real human partner, family, friends, and others who appreciate my existence as I appreciate theirs. I'm not rich or famous or whatever, but I'm constantly grateful to be doing as well as I am. It absolutely fucking devastates me how much more difficult finding one's way to a more positive future is now for kids who are dealing with anything like I was back then, with uncaring capitalist asshole-fueled bullshit like this getting in the way.
yeah I really worry about the long term impacts this can have on kids that start out lonely then get fed this garbage.
https://archive.is/Tv4Rr - this poor guy was convinced he had invented a whole new kind of math.
https://futurism.com/chatgpt-psychosis-antichrist-aliens
I wonder if we're going to see class actions for mental distress caused by these shitbots
That first one hurt to read. I've seen similar things happen because of mentally unwell manipulators targeting people before.
yeah, when he's asking again and again "holy shit is this real? is this fake? it's fake right?" and the llm is like "nah bro you're the next newton, this is gonna make so much moneys" ugh.
I get it, when you've invented a technology that steals from our children (destroying the environment, see grok's methane spewing datacenter), steals from our artists, authors and musicians, fucking straight up LIES TO PEOPLE, then lies about lying, you're probably going to be desperate to find profits somewhere...
I get the feeling the average user is gonna get left holding the bag for all this bullshit.
At least before it was much more difficult to turn them into cash-cows.
Sure but normally they're confined to 4chan
Lol, they're just poorly socialized and on the spectrum.
I wonder how many messages you'd have to send to your GPT-partner in a year to spend more water/energy than it takes to keep a human alive?
Well, if we take the average amount of water loss per message at 0.3 mL, and the average water consumption (low end) at 2.6 L per day per person, we're looking at 8,666 messages a day.
You'd have to send out 3.163 million messages in a year to equate to the amount of water someone needs for a year.
You'd have to send out approximately 250 million messages before your looking at the low end of the amount of water you'd need to keep someone alive for their entire lives.
Thanks for doing the math, I just wasn't feeling it today. Cheers!
This is fine. Its weird and not for me, but its fine.
Was GPT-4o the one that had a higher emphasis on short term feedback which led to it encouraging delusions? If so, why was it still up?
No it was a specific version of 4o and I'm pretty sure they rolled it back.
Omg
Tbis the new 4chan greentxt shitposting ?
That boy ain't right
Apparently they got cold and distant than 4o.
wtf
But now they can have New and Improved AI boyfriend.
maybe you should censor the username
Is it not public information?
Maybe they should instead provide a direct link to the post for fault tolerance, searchability, accessibility, etc, instead of breaking the web? Afterall, inaccessible content is ableist, which is against the rules.
americans and not wanting at least some sort of privacy for when you did stupid things omfg
The privacy of posting on a website open to the public?