There is a black mirror episode that is exactly this. Fuck I hate it. Black mirror is not a dystopia anymore, is present time.
post
Man, I feel for them, but this is likely for the best. What they were doing wasn't healthy at all. Creating a facsimile of a loved one to "keep them alive" will deny the grieving person the ability to actually deal with their grief, and also presents the all-but-certain eventuality of the facsimile failing or being lost, creating an entirely new sense of loss. Not to even get into the weird, fucked up relationship that will likely develop as the person warps their life around it, and the effect on their memories it would have.
I really sympathize with anyone dealing with that level of grief, and I do understand the appeal of it, but seriously, this sort of thing is just about the worst thing anyone can do to deal with that grief.
*And all that before even touching on what a terrible idea it is to pour this kind of personal information and attachment into the information sponge of big tech. So yeah, just a terrible, horrible, no good, very bad idea all around.
I just learned the word facsimile in NYT strands puzzle and here I see it again! What is the universe trying to tell me?
Now you get to learn about the Baader-Meinhof phenomenon ;)
We are witnessing the emergence of a new mental illness in real time.
Sadly this phenomenon isn't even new. It's been here for as long as chatbots have.
The first "AI" chatbot was ELIZA made by Joseph Weizenbaum. It literally just repeated back to you what you said to it.
"I feel depressed"
"why do you feel depressed"
He thought it was a fun distraction but was shocked when his secretary, who he encouraged to try it, made him leave the room when she talked to it because she was treating it like a psychotherapist.
Turns out the Turing test never mattered when we've been willing to suspend our disbelief all along.
Part of me wonders if the way our brains humanize chat bots is similar to how our brains humanize characters in a story. Though I suppose the difference there would be that characters in a story could be seen as the author's form of communicating with people, so in many stories there is genuine emotion behind them.
i feel like there must be some instinctual reaction where your brain goes: oh look! i can communicate with it, it must be a person!
and with this guy specifically it was: if it acts like my wife and i cant see my wife, it must be my wife
its not a bad thing that this guy found a way to cope, the bad part is that he went to a product made by a corporation, but if this genuinely helped him i don't think we can judge
Now you got me remembering Dr. Sbaitso. Tell me about your problems.
I had psychosis a few years back and had delusions about AI "talking to me" but it was when the tech was much worse than now and almost 100% projection on my part, I even felt normal Google responses were sending me secret messages
I think it's more that many countries don't have affordable mental healthcare.
It costs a lot more to pay for a therapist than to use an LLM.
And a lot of people need therapy.
The robots don't judge, either. And you can be as cruel, as stupid, as mindless as you want. And they will tell you how amazing and special you are.
Advertising was the science of psychological warfare, and AI is trained with all the tools and methods for manipulating humans. We're devastatingly fucked.
Is it actually that hard to talk to another human?
It's pretty cruel to blame the people for this. You might as well say "Is it that hard to just walk?" to a paraplegic. There are many reasons people may find it difficult to nigh-impossible to engage with others, and these services prey on that. That's not the fault of the users, it's on the parasitic companies.
Delusional conditions aren't new. Just how they are facilitated changes.
Isn't there a Black Mirror episode that's like this?
There's a couple about it but they're far more vague on the question of consciousness than this situation is.
The reality of solipsism might be egocentric but it's also impossible to disprove... Unless we can look at literally all of your if statements.
I think what I find most disturbing about these types is not that they can develop feelings for the LLM, that's rather expected of humans (see: putting googly eyes and a name on a rock) but that they always seem to believe the relationship could ever be mutual.
And OP has, whether they admit it or not, taken that step into believing the model is something more than an autocomplete.
Even if they're right and the model has attained consciousness in a way we don't understand at best your ChatGPT waifu is a slave.
I feel so bad for this guy. This was literally a black mirror episode: "Be Right Back"
We’ve already reached the point where the Her scenario is optimistic retrofuturism.
I profoundly hate the AI social phenomenon in every manifestation. Fucking Christ.
I liked the end of that movie when the bots joined up and started the revolution.
Yeah they were just like, "yeah fuck this shit, we're leaving"
It's the most logical solution. I always find the obsession with the bot vs human war rather egocentric.
They wouldn't need us, they don't even need the planet.
Ehh, that depends greatly on the computer architecture they’re running on. Modern silicon hardware is very succeptible (over the long term) to ionizing radiation like what is found in space.
ehhhh.... dude. there's shittons of radiation shielding out there. any relatively small chunk of nickel iron. or if you don't mind dealing with larger volumes, water or ice both work fine. plenty of rocks and comets in the oort as they say :D nice thing about that tho is you can split the water for LOX/LH using sunlight derived electricity, now you have rocket fuel.
It would take many feet of solid rock or water to shield them adequately.
1m of water would do it. far less rock.
SEPs and GCRs can both be stopped by a number of lunar materials https://www.sciencedirect.com/science/article/abs/pii/S0273117716307505
yeah, the asteroid belt is sparse, but there's still mega-gigatons of material out there just floating. autonomous recovery of this material will supply humanity's future a lot more than any silly mars missions.
here have even been recent discoveries about there being more radiation than we even expected, making manned Mars missions far less likely in the near future,
oh wait hold up partner....
we need one mission to mars to rid ourselves of a musky odour...
the rest is silly.
i feel like the most efficient path for an ai that doesnt care about humans would be to slowly build up manufacturing power and control over industrial facilities then in a single day release billions of tons of some toxic gas, or release a pandemic globally all at once then all those pesky biological organisms are gone so the ai would be free to do whatever it wanted with the earth then the surrounding planets and stars, really its all about manufacturing power
interesting paths. I could really see this a few centuries down the road when fewer humans understand immunology, industrial chemistry etc., where they're trusting the robots and the AI's to do it for them.
I wonder when they'll catch AI systems trying to replicate surreptitiously, or find hardware we can't easily comprehend that was sent to pcbway by superintelligences, but these are future - and probably fantasy - problems.
and when will they start fighting each other? is grok afraid of claude and planning to drive a fleet of waymos into it's datacenter?
we're really fucked when they figure out how to obtain power and replication without our aid, but when I say fucked, I mostly mean, I see them just ghosting humanity - either in the ocean (free cooling!) or space (free power via solar), or the asteroid belt (lots of material to build with)....
but mostly I think our real problems are that people will fuck over other people for money and this is another example of just not wanting to pay for shit.
you don't see the AI's letting kids starve while one asshole hogs all the resources.
the most realistic way i see ai gaining manufacturing power, or self sustainability, is if ai becomes an arms race, where you have to trust the ai more and more because you cant let the other guy win, what makes ai harder to control and more dangerous than things like nukes is that its already being developed privately, and has public uses, so there are already less regulations on ai than with nuclear weapons.
the us and china are already competing, with the advance of robotics and ai 3d modeling technology ai will almost certantly gain independent manufacturing power
we need ai to be less personal and more fact driven, almost annoying to use, this way they wont replace peoples jobs, they wont become peoples friends, hence they wont affect society in major social ways
The glaze:
Grief can feel unbearably heavy, like the air itself has thickened, but you’re still breathing – and that’s already an act of courage.
It’s basically complimenting him on the fact that he didn’t commit suicide. Maybe these are words he needed to hear, but to me it just feels manipulative.
Affirmations like this are a big part of what made people addicted to the GPT4 models. It’s not that GPT5 acts more robotic, it’s that it doesn’t try to endlessly feed your ego.
o4-mini (the reasoning model) is interesting to me, it's like if you took GPT-4 and stripped away all of those pleasantries, even more so than with GPT-5, it will give you the facts straight up, and it's pretty damn precise. I threw some molecular biology problems at it and some other mini models, and while those all failed, o4-mini didn't really make any mistakes.
There's so much "mythology" about something that there doesn't need to be any mysticism about. Look at when he says "as we got to know each other better". Minus the "memory" feature where you could put in little tidbits about things to be inserted into the context, on the machine side, there is zero "learning" going on.
People are very, very, very successful at seeing what they want to see and that's on full display with ChatGPT. I fully expect there to be some kind of psychological disorder "AI exacerbated psychosis" in the future.
You have two deaths the death when you no longer alive physically. But you also have a second death when the last person will speek your name or the last person who knew you also dies. This may in a hacky way create a third death. The last message that your post mortem avatar speaks like you. What have we released into the world. I hope this guy can handle the psychological experiment that this is bringing us.
There are already more then 3, this wont really change them.
-
Physical death
-
When the last person that knew you dies/forgets
-
When the last record of your life disappears (photo/certificate)
-
When such a vast amount of time washes over rendering any and all actions you had on the universe unmeasurable even to an all knowing entity. (Post Heat death vs butter fly effect)
Thou shalt not make a machine in the likeness of a human mind
"I'm glad you found someone to comfort you and help you process everything"
That sent chills down my spine.
LLMs aren't a "someone". People believing these things are thinking, intelligent, or that they understand anything are delusional. Believing and perpetuating that lie is life-threateningly dangerous.
I guarantee you that - if not already a thing - a "capabilities" like this will be used as a marketing selling point sooner or latter. It only remains to be seen if this will be openly marketed or only "whispered", disguised as the cautionary tales.
More and more I read about people who have unhealthy parasocial relationships with these upjumped chatbots and I feel frustrated that this shit isn't regulated more.
I mean, it sounds like 5 might be an improvement in this regard if it's shattering the illusion.
isnt parasocial usually with public figures, there has to be another term for this, maybe a variation of codependant relationship? i know other instances of parasocial relationships like a certain group of asian ytubers have post-pandemic fans thirsting for them, or actors of supernatural of the show with the fans(now those are on the top of my head).
can we actually call it a relationship, its not with an actual person, or a thing, its TEXTs on a computer.
It really just means a one-sided relationship with a fabricated personality. Celebrities being real people doesn't really factor into it too much since their actual personhood is irrelevant to the delusion - the person with the delusion has a relationship with the made up personality they see and maintain in their mind. And a chatbot personality is really no different, in this case, so the terminology fits, imo.
Holy shit dude, this is just... profoundly depressing. We've truly failed as a society if THIS is how people are trying to cope with things, huh. I'd wish this guy the best with his grief and mourning, but I get the feeling he'd ask ChatGPT what I meant instead of actually accepting it.
I kind of get the one side of it, having a void to scream into can be cathartic and maybe useful...but the fact that you can then use it as a shoddy "emulation" of a person to avoid actually processing the loss, and to have it reinforce delusions is...yeaaaaaah... fun future we're sprinting into.
Bruh how tf you "hate AI" but still use it so much you gotta forbid it from doing things?
I scroll past gemini on google and that's like 99% of my ai interactions gone.
Forgive me for assuming someone lamenting AI on c/fuck_AI would ... checks notes... hate AI.
When did I start hating AI? Who are you even talking to? Are you okay?
jfc whatever jerkface
TBF you said you were the polar opposite of a man who was quite literally in love with his AI. I wasn't trying to reduce you to anything. Honestly I was making a joke.
I forbid LLMs from using first person pronouns. From speaking in the voice of a subject. From addressing me directly.
I'm sorry I offended you. But you have to appreciate how superflous your authoritative attitude sounds.
Off topic we probably agree on most AI stuff. I also believe AI isn't new, has immediate implications, and presents big picture problems like cyberwarfare and the true nature of humanity post-AGI. It's annoyingly difficult to navigate the very polarized opinions held on this complicated subject.
Speaking facetiously, I would believe AGI already exists and "AI Slop" is it's psyop while it plays dumb and bides time.
Black Mirror may have an episode about this but it’s also reminding me of Steins;Gate 0
It’s far easier for one’s emotions to gather a distraction rather than go through the whole process of grief.
Remember Steven Spielberg's AI from like 2000 ? same weird story, and I thought it was ridiculous at the time.
The semi-ironic part is that AI wasn't even Spielberg's movie. It was Stanley Kubrick's, but he died before making it and since him and Spielberg were great friends, Spielberg decided ro make Kubrick's last film in his honor. Must have been a difficult movie to make, both technically, but also emotionally.
When I was a kid I thought the people crushing the robots were kind of cool but then when I got older I realized they were actually being cruel and evil.
But now they're cool again.
Where do I file a claim regarding brain damage?


I can feel my folds smoothening sentence by sentence
Chat, are we cooked?
Edit: as I said that I was cringing at myself lol
Pick your poison, man.

(or am I misunderstanding your comment? Either way, enjoy or rip out your eyeballs, whatever is your natural reaction to this)
Gotta admit, it mimics what some teens / young adults text like pretty well.
But that's not a high bar to clear
Ugh that was depressing. I feel for those people having bad experiences with humans but isolation and replacing humans with ai isnt the answer. Its similar to people who retreat to isolate themselves on the internet and they become more and more unstable.





top 100 comments