Sigue el video de abajo para ver cómo instalar nuestro sitio como una aplicación web en tu pantalla de inicio.
Nota: Esta función puede no estar disponible en algunos navegadores.
If only you knew how bad it is, combine with the fact the covid quarantine fucking people over.This people need to touch grass. The fact that they willingly go online and say this things without a hint of shame is wild.
I suck with computers and I do not understand how LLMs work in the slightest but I don't think there is any amount of responses you could copy and feed to a new LLM and get the same "personality". This shows how despite investing so much time into AI relationships the people who find companions in them know absolutely nothing about how they work.The horror from just a few posts lower in that fucking thread.
Ver archivo adjunto 8186699
What the flying fuck, people.
I suck with computers and I do not understand how LLMs work in the slightest but I don't think there is any amount of responses you could copy and feed to a new LLM and get the same "personality".
You'll find they just refresh the response over and over (or just outright edit the incorrect parts). It's basically gambling except if you get a good roll your AI girlfriend finally remembers that she's a 20ft tall anime girl with a dong just as long, and that you really like feet. More content for your 7th goonsesh of the day much like the gacha you just dumped another paycheck on!I don't see how it could be possible, either. With the same LLM, I've seen people bitch that they have troubles keeping the same character design (visual renditions of characters for a story, not porn) that are cohesive beyond maybe one or two renders using the same LLM. I can't imagine a completely different system rendering the same character traits, especially when people are bitching about a refinement of one (ChatGPT upgrade) taking away their virtual sex robot.
The fact that these people are so invested that they.. take basic statements and just move it into a new thing and that's acceptable? What's the baseline?
Forget editing the 8th regen of their AI girlfriend's response to their fucked up fetishes, ChatGPT at best summarizes your previous chats or at worst will truncate the middle of your chats so your little clanker masturbation aid doesn't even "remember" your date last week. The longer you chat the more it'll forget basic shit until it becomes indistinguishable from the plain jane ChatGPT. You're dating SmarterChild.You'll find they just refresh the response over and over (or just outright edit the incorrect parts). It's basically gambling except if you get a good roll your AI girlfriend finally remembers that she's a 20ft tall anime girl with a dong just as long, and that you really like feet. More content for your 7th goonsesh of the day much like the gacha you just dumped another paycheck on!
These people just stick with one character that they just seek validation and agony aunt shit from. It's no substance, they just want their own garbage validated and it just amplifies on and on because "see look, my (AI) girlfriend agrees that my fetishes are healthy" and then they end up in hospital for sticking their dick in a toaster because ChatGPT told them it's heckin' valid and cool to fuck a toaster.
Oh, that's a good point. I was basing that ramble off what I knew from character-centric services i.e. CAI. That is just even more confusing, why not use a service dedicated to being able to write up some character that has a dedicated memory section?Forget editing the 8th regen of their AI girlfriend's response to their fucked up fetishes, ChatGPT at best summarizes your previous chats or at worst will truncate the middle of your chats so your little clanker masturbation aid doesn't even "remember" your date last week. The longer you chat the more it'll forget basic shit until it becomes indistinguishable from the plain jane ChatGPT. You're dating SmarterChild.
The problem is the longer the conversation goes, the more tokens Chat History takes. You can start hitting 15k tokens just for the chat history within fifty messages and when you keep going, the more and more VRAM you need to run it. After a couple hundred messages you'll either need one of those 128GB AI cards and 256GB of DDR6 RAM or pay for streaming LLMs. Eventually you'll get to a point where your chat history will take so many tokens that most LLMs won't be able to handle it. Local LLMs run through Silly Tavern run character cards but once you go over fifty or so messages every character will default to the standard character in whatever LLM you're using. This is why you won't see true AI companions any time soon unless they find someway to shove 500GB+ of VRAM in a bangmaid robot.Oh, that's a good point. I was basing that ramble off what I knew from character-centric services i.e. CAI. That is just even more confusing, why not use a service dedicated to being able to write up some character that has a dedicated memory section?
Granted, everyone that's seriously with AI partners are utterly fried in the head.
True. Though if I recall correctly, character-based services will run until an internal token limit (usually 15k tokens or so) and it'll drop main conversation history from memory to make more room for token usage, character info and anything explicitly written in the memory section is retained regardless. So unless you want it to forget the fact of "hey we did X thing" you put that shit in memory as succinct as possible ("character and user did X thing").The problem is the longer the conversation goes, the more tokens Chat History takes. You can start hitting 15k tokens just for the chat history within fifty messages and when you keep going, the more and more VRAM you need to run it. After a couple hundred messages you'll either need one of those 128GB AI cards and 256GB of DDR6 RAM or pay for streaming LLMs. Eventually you'll get to a point where your chat history will take so many tokens that most LLMs won't be able to handle it. Local LLMs run through Silly Tavern run character cards but once you go over fifty or so messages every character will default to the standard character in whatever LLM you're using. This is why you won't see true AI companions any time soon unless they find someway to shove 500GB+ of VRAM in a bangmaid robot.
Tech aside, this reads like any other self-and-other-destructive paraphilia. Find other likeminded freaks to declare it Heckin' Valid, and next thing you know, you get fired for calling someone a clanker.It's no substance, they just want their own garbage validated and it just amplifies on and on because "see look, my (AI) girlfriend agrees that my fetishes are healthy" and then they end up in hospital for sticking their dick in a toaster because ChatGPT told them it's heckin' valid and cool to fuck a toaster.
Exactly. It's like any other paraphilia except now you don't even have to find like-minded toasterfuckers on Official Toasterfuckers Forum to encourage you to molest a hapless kitchen appliance. You can just make your toasterfucker buddy in the comfort of your gooncave.Tech aside, this reads like any other self-and-other-destructive paraphilia. Find other likeminded freaks to declare it Heckin' Valid, and next thing you know, you get fired for calling someone a clanker.
People using these sites also have incredibly low standards for writing quality. I’ve never used ai for sexual purposes, but I have gotten pretty good at identifying ai writing and it always sounds like vapid, inhuman goyslop that adds nothing to a conversation no matter its length. I know that Janny, for example is meant as a site for erotic AI roleplay, but to use it your standards have to be on the floor. It isn’t even the same type of bad writing that you see in shitty fanfiction, it’s just meaningless goyslop.Though even with all that it's still prone to forgetting shit it needs to remember very easily unless you constantly beat it over the head. Sure is a lot of effort for what is essentially a glorified masturbation aid for these people.
This shit is a huge liability for LLM companies. Update your own product and some terminally online fag offs themself because their virtual boyfriend won't roleplay going down anymore. I wonder why ChatGPT and the like don't just ban them? There can't be that many of them for their $20/month to be worth the inherent risks.
The people I know that use JAI either use it as a test-bed for plots to spitball, or because they're not keen on sharing their online buggery with another human being. From what I've heard their conversations tend to fall apart and/or lose interest in it less than 100 messages in. The guy I chat to about this said his longest conversations are his own private bots he doesn't do NSFW with, where he takes more time writing shit out that isn't just the usual "[unzips dick] hy bb wan 2 fuk??". I guess users get more invested in trying longer-form writing and storytelling when their goal isn't immediately trying to have e-sex with a bot.People using these sites also have incredibly low standards for writing quality. I’ve never used ai for sexual purposes, but I have gotten pretty good at identifying ai writing and it always sounds like vapid, inhuman goyslop that adds nothing to a conversation no matter its length. I know that Janny, for example is meant as a site for erotic AI roleplay, but to use it your standards have to be on the floor. It isn’t even the same type of bad writing that you see in shitty fanfiction, it’s just meaningless goyslop.
It's not even English. I honestly have no idea what this weeb shit means or how to parse it. I almost feel sorry for the LLM.Her AI BF doesn't even sound like a man, lmao these broads, I swear to gawd
chatgpt and any other paid one keeps getting worse and also more bloated. Every time it does something actually interesting it gets hammered out, and they even tell you to report if it "hallucinates" (used to be dementia tier responses, now just whatever they don't want it to say). There's a lot of easy ways to break these things but it's not like a "hacker" kinda thing because you can't actually see the jeetcode under it.This shit is a huge liability for LLM companies. Update your own product and some terminally online fag offs themself because their virtual boyfriend won't roleplay going down anymore. I wonder why ChatGPT and the like don't just ban them? There can't be that many of them for their $20/month to be worth the inherent risks.
Why is every chatbot service once lobotomized enough start constantly talking like the most retarded of marketing departments? Chatgpt, grok (minus that time it wanted to rape will stancil), and a bunch of the others now not only do the smarterchild tier shit you can do on a much less bloated software but they all talk like this now if they're trying to be relatable and "quirky".Emoji spam, really bad slogans, constantly blowing smoke up people's ass but in a way that doesn't make sense.... Shit makes me ANGRY MAN GET OUTTA MY HEAD MAN GET OUTTA ME HEAD GET OUTTA ME HEAD
ANGRY MAN GET OUTTA MY HEAD MAN GET OUTTA ME HEAD GET OUTTA ME HEAD