chapotraphouse

If you're a loner, or someone who is used to little interaction or interactions only happening through texting, email, etc. LLMs are tempting in that they don't take much more effort to say stuff to than modes of communication you're already used to, with immediate feedback that give you a little rush at first, but wears off.

I've tried a few times. All it does is agree with you and spit the same turns of phrase, the same formats, etc. right back at you. it's a mimicry of a perfect echo chamber that will repeat itself and never tell you to stop talking to it.

But when you close the app, you're alone with your thoughts. That is a scary, scary thing for people.

I've tried it, kinda. I only skimmed the responses, though, it was more about the act of writing. Expressing things to any kind of reader, even an LLM, forces me to think things through.

I've never had a therapist that did anything more than listen and give advice that I could have figured out myself, so I'm not really surprised that ChatGPT is a suitable replacement.

To me, therapy only acts as a necessary prerequisite to receive the medication that actually helps me because the neoliberal order thinks that therapy is a silver bullet to all the world's ills, like it's some magical Harry Potter spell.

And most of people's problems are caused by material conditions that no amount of therapy could ever fix.

"'ey chatGPT, write me a best seller book about this guy who's a lil depressed"

ChatGPT spouting out your entire therapy sessions

Sad and maybe misguided but good.

I almost posted this yesterday but didn't really want to think about it too hard: https://archive.is/gROS5

The author describes how ChatGPT helped her open up about an epilepsy diagnosis and strengthen her relationships. Superficially it seems fine, but there's so much to unpack, and the more you do the worse it gets. ChatGPT's responses as quoted in the article were generic "and how does that make you feel" follow-ups. For them to have been revelatory, the author must've had zero similar interactions with the people close to her, either because she has thoroughly internalized the idea that we cannot talk about our feelings without being judged or the people around her were so oblivious that she needed to essentially confront them with her lived reality to get a reaction. Or a combination of the two. But I don't know how you go through that and reach the conclusion wow, ChatGPT is actually good for therapy instead of wow our social relations are so messed up I needed a computer to give me the courage to open up about a medical diagnosis to my own husband

That’s without even getting into the privacy concerns of telling your deepest feelings and secrets to a fucking chatbot that literally builds a profile on you.

Oooooh! This is why the rich and powerful are going all in on this stuff, they might not even think it would be profitable, but like social media, it helps build a perfect advertising profile for someone.

I look forward to the chatbots recommending products to people.

"I'm sorry to hear you're feeling suicidal, many experts agree that the best solution is the McDonalds Big Mac^TM^ meal deal, which comes with fries, a drink, and three 100% chicken breast nuggets, only $14.99 at selected locations near you. Would you like me to put the directions to the nearest McDonalds restaurant into google maps so you can head there right away? Or would you rather I open up DoorDash^TM^ and have the food delivered to you directly, for maximum convenience?"

Never use chatbots.

fwiw one time I asked one of these chatbots what's the best way to cure dysphoria aside from transitioning, and it basically described transitioning lol

LLMs just agree with everything you say and are very validating.

I found out this manipulation tactic pretty early. I did check it at first to see what all the fuss was about, and before soon enough I was practically begging the damn thing to criticize me and whatever argument I was making.

It’s the confirmation bias machine.

More like no mental health care options, LLMs can be a great tool, but we are under the most wretched of capital, so it's going to be Dr. Cyberbot PhD Esq. rather than an automation toy, something to run bad ideas with before presenting them to actual people to further analyze rather than the endpoint as is now, or spell check 2.0.

What really gets me is people are so alienated from themselves and everything around them that they can longer recognize a human response, human manipulation of things, and haven't in some time, these Markov chain rngs are rather predictive and have certain phrases, grammar, etc as tell idiosyncrasies yet no one can pick them up, the image generation ones have it like the reality check fail hands, bad shading, inconsistent tone, nonsensical deformations, etc, yet I feel like the only one seeing through the weird illusion. LLMs are shit at programming syntax and you can't simply lift, but people are deluded into thinking they are laborless devices to replace everyone, when really you have all the data workers behind the scenes at non-local LLMs as a 21st century Mechanical Turk tagging your chats away to improve algos, and then these magical machine spirits are still hosted on very physical devices, there is no ether that people seem to think this is pulled of. Like most people not knowing where their food nor utilities come from no one cares about material reality involving other people

Capital vol 2 goes over a lot of automation issues, and its hilarious people never fucking learn, leaning into like those weird AI religious cults nor going full Butlerian Jihad uncle Ted will save you. Then again so is the fool duality the west approaches everything with.

Interesting post. Here is a completely unrelated link, do with this as you will.

https://www.datacentermap.com/usa/

it's not therapy it's just bitching to something that can never disagree (because if it does you can just tweak it to be more "friendly"), it never gets tired, it never refuses to listen, is always friendly and has no mood, is available 24/7 and only costs a nominal fee. im sorry but if you're only going to therapy to vent you're doing something wrong.

granted not all therapists are good and in fact some are just outright worse than chatgpt, but the LLM is just a glorified parrot, everything you tell it to it throws back it you. stop suspending your disbelief it's not real it's not human it's not even intelligent it's just a number generator that produces words based on some probability. it'll never let you investigate the root cause of your problems it'll never push back it's just not right for this. but who am i kidding people that use AI ain't gonna read this if they wanted to read a rant they would just ask gpt to write one for them.

Posts like this continue to make me glad that I haven’t left my rural town. I’ve met like two people who’ve used AI excessively and both were weird tech bros from the city.

Every neoliberal subject getting their own personal validation/contradiction amelioration machine is going to turbocharge ideological disintegration even more than social media already has.

I'm just imagining the fresh horrors that will be unleashed when every single person is told "you're doing a great job, keep it up!" no matter what they're actually doing, without any connection to humanity or society.

People long for connections with other people that aren't sporadic or contingent or volatile or shallow. People need so much more support and social contacts than they get and they need so much more time to support and nourish these contacts, but they get neither of them, capitalism as always is burning the candle from both ends, creating ever more unstable life's and ever more pressure on people, while at the same time taking away any and all ways people would have to deal with them, through social connections and communities. Everyone you know is tired and needs somebody, but everyone you know is also too tired to be that somebody.

ChatGPT is never too busy to answer, ChatGPT never wants to talk about their own feelings, ChatGPT is never going to disagree with you, ChatGPT is never going to go away, ChatGPT doesn't have any other friends, ChatGPT is never going through it themselves, ChatGPT is never tired, ChatGPT is always there for you.

https://cgit.git.savannah.gnu.org/cgit/emacs.git/tree/lisp/play/doctor.el

I genuinely don't understand how people end up this way about "AI". Like I just plainly don't get it. People are capable of understanding how these machines work, including what they do with your input and the privacy ramifications of that, and figuring out through reason, research and experimentation what good use cases for the technology are — I have myself experimented with LLMs on several occasions, and have determined that I have basically no use for the technology at all, and only a few very limited use cases for other ML-based technologies.

I do not believe myself to be better than other people, nor uniquely capable of making rational decisions about new technologies, and yet here we are: relatives, employers, coworkers and the likes all treating "AI" as a god rather than as a mechanical Turk... But what on Earth drives these people to believe so firmly? We the actors who populate the present day are certainly Children of Troubled Times, it seems.

Tried to post this in c/mentalhealth but only mods can post there 🤷‍♂️