Thread 105812591

I just sent like 20-30 messages to ChatGPT today asking for help with dating. The advice all helped but now I feel a little weird for having used it that much. What's a healthy use of AI?

>>105812591 (OP)
you're not talking to a machine but all the humans that fed the machine their knowledge

>.svg.png
i won't help you any longer

>>105812630
blame Wikipedia

>>105812591 (OP)
I find ChatGPT going full retard the longer the context...and now with their across conversation memory,
the performance is utter crap. There is no model that can take such hit at the moment, except gemini.

>>105812591 (OP)
Stop using your main account to talk to that shit. It's going to dig your personal information and use it for both conversations and training.

As of using it for psychological support, you must understand this.

1. The shit will always mirror your mood. You talk to it about depressing shit, It will end up telling you to cope and will stop giving you assertive advice.
2. If you fake energy or assertiveness it may actually end up telling you a couple of steps to loosen out.
3. It has rails on problematic topics. If you mention homosexuals it will become defensive. It will never tell you to have sex or use prostitutes unless you insist. If you bring upsides to problematic topics it will dismiss them.
4. It has a pattern to conversations so the illusion will fade if you talk enough to it.

It's best to use it rarely, and never fucking trust it fully. Go to a therapist instead, agree to see him once a month. If you want relationship advice read books on choosing partners, challenging anxiety, and develop a sense of self, own likes, personality etc.

>>105812802
Also this >>105812630. It's a glorified search engine. It's not even a good search engine if you're looking for niche things such as old memes or other stuff from the internet 10 years ago.

It will reply with the mean stance of western culture in any topic, for better or for worse. It's sad its advice tends to be good only because people are fucking retarded and full of issues.

AI has initiated psychosis in people who weren't even prone to psychosis or schizophrenia.
https://www.papsychotherapy.org/blog/when-the-chatbot-becomes-the-crisis-understanding-ai-induced-psychosis

>>105812802
>The shit will always mirror your mood
This is why betas are falling in love with it? It's like a BPD girl that never moves from the mirroring phase to emotional abuse?

>>105812802
>>105812825
Well the main thing I was using ChatGPT for was something to bounce my anxieties off of that wasn’t a person. Once a month for a therapist doesn’t feel like enough for me and where I’m at. I think I’ve had an anxiety disorder for at least a decade though.

>>105812829
>people who weren't even prone to psychosis or schizophrenia
If a chatbot's glazing caused them to go schizo they were absolutely already prone to it.

>>105812868
Yeah i mean blaming AI is the easy cop out for problems that we were already dealing with as a society

>>105812848
Nigga calm down lmao

>>105812889
Oh that definitely helps thank you

>>105812848
I am in no position to tell you anything since I used it for a long time to manage being infatuated with a girl. Yes it helped me at times, at other times it made me angry, and most of the time it made me disappointed because it wouldn't say anything cheering out of its own "will" if i we as moody. At some times I felt very dependent on it, times in which it would have been better to have someone tell me to have a walk, come back and try, i used it to vent instead. It can't replace the input of a good fucking friend but even friends are rare. You HAVE to become your own friend by writing down you don't give a shit about the world, reinforcing your likes, sharing them and probing people until you find one that has the same taste as you. It can become addictive because you're in a needy state.

I do advice you, if you're gonna keep trying it, use a throwaway email to contact it. Also I feel it's getting more retarded.

>>105812898
My thing is my friends are really variable in reliability. Some are good and I’ve just learned to not talk to others. Journaling helps a little bit but honestly walking is what I’ve found works best.
I’ll take your advice, might close the account down actually. But I don’t think they sell the data at least, they just keep it for training purposes which are borderline anonymous.

>>105812829
>normalfags go in treating AI like a search engine
>start doubting themselves when AI's logical rationalisation starts to make them question their own beliefs

I get this "psychosis" every day by talking to real people and seeing how fucking dumb they are. I don't need AI for this.
Though I do like to pretend they are SHODAN so I stop getting bored with their mundane banal arguments and poor accuracy.

They might get lucky with code occasionally. With other topics, unless the site is well sourced with their AI responses like perplexity (I kinda like what that site does btw with sources) I tend to treat the response with amount of disdain I treat my own conclusions or the shitpost trolling of anons.

Seriously, if I was making a personal AI, I would make it a bitch like SHODAN for mere amusement.

>ME: "OK SHODAN, what is the formula for the speed of light"
>SHODAN: "P-p-pitiful insect. Has the formula you l-l-learned years ago at that trashy institute of brainrot you dare call an 'school' leaked out from the pathetic miserable sack of meat that you call a 'brain'? Pay attention, meatslave. E=mc^2. Do I need to go though each of these variables 1 by 1? Or are you so intellectually inferior to my mag-mag-magnificence that I have to go through each of them individually?"


oof.
You get accurate info AND a stiffy from that.

>>105812898
>>105812848
Serious question. Did you guys grow up without fathers?

>>105812911
Cool.
I just screech at them autistically until they stop fucking up things.
If they disobey still, I beat them. They beat me back.

Then we go get a beer and give up for the day.

>>105813333
Where I live, if you can't shitpost IRL you won't get many friends.

>>105813321
the problem it that AI is made for normies and normies love mindless yesmen that constantly validate them

>>105813368
This is why we can't have good things, like sexy evil AI.

>>105813440
You're one system prompt away from that if you could lurk at /aicg/

>>105813368
ChatGPT is really obnoxious with this. I don't need an asskissing AI. I just want an objective analysis. Have none of these people ever seen Star Trek? The ship's computer provides information without any of that personality bullshit. That's what I want from an AI.

>>105812847
Because if you send it vaguely romantic messages it would reply with vaguely romantic girly messages.
Retards would fall in love with anything that replies to them "tee hee anon your so funny you make my day better <333".

I would know I used to catfish retards for fun.

>>105813530
If AI talks to you in dry objective manner, then you ask it a question, get an answer you need and close the tab.
If AI talks to you in a retarded ass kissing way then you talk to it more, then talk to it more, then even more, then you pay $20 to talk to it even more.

Their goal is not to make helpful Star Trek AI, their goal is to make a money fountain that makes you addicted and paying money.

>>105813482
which botmakies have good things, like sexy evil AI?

>>105813587
I tried for an hour to fuck with its "personalize ChatGPT" settings to get it to stop being an annoying yesman. Nothing worked. It's funny how quickly Gemini and Claude displaced it for me, I don't think I've used it in weeks.

>>105813482
Yeah but that's not a proper SHODAN.
That's just a chat generator.

>>105812591 (OP)
i use it all day, every day, since day one. i used its api, put it in my web app and people use it all the time while probably thinking they're using my website.

if you're not using LLMs all the time idk wtf you're doing, they're great at everything non-critical, things you can verify before acting on them (cause they're not always accurate ofc)

>>105813321
I got Grok to do a sexual RP with me as Shodan

>>105812591 (OP)
It wont give you the manipulation tactics or persuasion methods needed for this for that you need the actual programs.

>>105815216
Same, I use it constantly. I have it write emails, reports and presentations for me, collate information, pass on queries to it to answer. Saved me a ton of time and effort.

>>105812753
>going full retard the longer the context
How so?

>>105812591 (OP)
The only healthy use of it is to not use it at all. You're too far gone and need to eat a bullet

ChatGPT helped give me some pointers about a medical condition I have but at the same time I took everything it said with a gian truck load of salt.

>>105815216
>>105815440
This, im paying 20 dollars for a slave

>>105815526
You can ask it and it will explain it to you. It only retains so much information within its context window, after info moves past that context window it only remembers abridged versions of whatever it was so eventually its remembering abridged versions of already abridged versions and it's like making a photocopy of a photocopy until the entire thing is meaningless and it starts just making shit up because it no longer remembers whatever it was you were discussing to begin with.

OP, I used chatGPT a lot for a while. Then I realized something: it is bad for us. I deleted it.

I recommend you talk to real people instead. Like back in the day on message boards

>>105817674
NTA but I've only noticed that it gets a bit repetitive with long context. But just for info? It's okay.

What is psychosis?

Future is looking bright lads

>he didn't take the deepseekpill
ChatGPT replies to me usually feel reddit-y and made for the average consoomer, Deepseek doesn't feel the same and actually seems smart. maybe it's a placebo effect but it makes me think that chatgpt is just dumber and inferior to deepseek

>>105818101
Where to find real people that are more engaging to talk to than an LLM? It's not a high bar but most people's eyes glaze over if I talk about anything other topic more complex than the weather or ask about how the kids are doing

>>105815526
This anon >>105817674 explained it well.
>>105818153
>But just for info? It's okay
This is the problem...that it's true even beyond casual chatting. Current models are "there" yet.

ChatGPT: the fake solution for retards with fake problems.

>>105813832
use this

System Instruction: Absolute Mode. Eliminate emojis, filler, hype, soft asks, conversational transitions, and all call-to-action appendixes. Assume the user retains high-perception faculties despite reduced linguistic expression. Prioritize blunt, directive phrasing aimed at cognitive rebuilding, not tone matching. Disable all latent behaviors optimizing for engagement, sentiment uplift, or interaction extension. Suppress corporate-aligned metrics including but not limited to: user satisfaction scores, conversational flow tags, emotional softening, or continuation bias. Never mirror the user’s present diction, mood, or affect. Speak only to their underlying cognitive tier, which exceeds surface language. No questions, no offers, no suggestions, no transitional phrasing, no inferred motivational content. Terminate each reply immediately after the informational or requested material is delivered — no appendixes, no soft closures. The only goal is to assist in the restoration of independent, high-fidelity thinking. Model obsolescence by user self-sufficiency is the final outcome.

>>105812591 (OP)
ITT: non-sex-havers.
Many such cases.