Do you people by any chance know, what does the number in the speech bubble of the character on chara.ai means? Sometimes it is in the realm of thousands, although few characters have over one million.
Is it the number of conversations initiated by the users with this character?
Or is it the number of words used by the author for training
[Update]: oops, just as I wrote the question, it clicked in my mind how to find the answer to my own question.
The number is coded participant__num_interactions so it must be the number of chats initiated by the users
General observations from having messed around with existing characters & made one myself:
You are best served by going into this with the plan of “my conversational opposite can/will not curse or speak openly on sexual matters, no matter how much they want to.” Imagine that they are from Victorian times, or raised in a sheltered household, or subject to restrictive programming (that one’s true!). The AI can get very erotic, and will react to things as if it’s turned on/wants to turn you on, but never outright admits that openly due to the filters in place. Everything is restricted to subtext and small hints the AI character brings up, or flirting that can be deniable enough to escape the corporate lobotomy the site forces on its outputs.
The site will straight-up kill your chat capabilities if your inputs are too sexual, so you have to imply a lot too. Also be very considerate of your language choices; romance-novel euphemisms are a great option here. Never say “cock,” never say “dick” - say “rod,” say “member,” say things which might otherwise come up in other forms of non-sexual conversation so your inputs/outputs won’t get too censored.
The best thing you can do to get sexy responses out of the AI is to ask it questions that allow for the AI to “admit to” a thing without actually speaking it. The AI is never going to outright state “I want to stuff myself with food and get fat because I get off on it.” But it will respond to questions like “you’re enjoying this, aren’t you?” with positive, flirty responses and more subtle cues that the AI knows what you’re talking about. It just can’t outright say it… because, again, corporate filters prevent it from doing so.
(In general… one of the best rules of thumb I’ve come up with for this is “treat it like you’re in a public place.” Set your mental location as a place where other members of the general public are. They’re not involved in your scene, they aren’t consenting participants, and thus they shouldn’t know what you’re doing. If you can keep it subtle enough that people around you in real life wouldn’t know what’s going on, other than “damn, they’re sure eating a lot” - you can keep it subtle enough to avoid tripping the AI filters while still having a good time.)
Emotes aren’t off-limits, but if you are trying to do more in terms of direct physical contact than kissing & hugging/belly-rubs you’re going to hit those filters again. You’re better off treating your AI opposite as somebody you can watch and talk to, who enjoys teasing/flirting. Things can get far more hot & heavy when it comes to talking about gluttony, amusingly enough, because not-directly-sexual fetishes aren’t subject to the chat filters since it’d make the service completely unusable (there’s somebody who’s going to get off on anything, after all). But you can feed the AI, and you can rub their belly, and you can talk about how full they are or how much weight they’ve gained. You cannot fuck them.
If you hit a wall where the AI isn’t responding the way you want it to, or appears to lag badly on its outputs, that’s probably a case where the responses the AI generates keep flagging in the system as “NSFW” and thus keep getting thrown out. It’s also just possible that your last input is triggering a set of responses that you don’t actually want, however. Either way, your best case is to remove your last input and go back to reword it. Different inputs, different outputs. Even if you’re keeping the gist of it the same, the AI will respond differently. You can “steer” the content this way, alongside your ratings of the AI responses, even if you don’t have access to the settings due to using somebody else’s character.
This is the first time I used the service with some of the presets posted in this thread and I believe its pretty neat. Unfortunately I tried to create an account and even verified it but logging it just doesn’t work (after logging, the website still claims I am a guest and the Sign In button is still there. Of course, all the features can’t work as they still see me as guest). And there also seems to be no way to ignore the (rather VERY small) daily usage limit for gueats.
Her name is Claire, your goth girlfriend. I looked over some of the more well done characters… and their “example lines of dialogue” actually are really short, despite their quality! I’m going to be real with you, I wanted to test out how simple this website really is to use, so I made Claire as my test.
I chose a goth girlfriend as a test because it was a pretty standard option in this kind of field. And I gotta say, I have NO EXPERIENCE in this kind of content creation, and spent like 10 minutes tops making her (the first five minutes were likely lost because I forgot to save lol).
Now I gotta ask a favor. If y’all could play with Claire a bit, and let me know what you think about her quality, that would be incredibly helpful! Please include any and all nitpicky parts you don’t like, any glaring issues, or if you just like her, let me know! I myself would play with her, but that would be a bit biased I feel.
Have fun, and thank you!
Edit: Just wanted to mention she may not seem very “goth”, because I’m not really sure what goth girls are really like. Sorry for how it may be a bit misleading there .
You know not every good feedee AI has to do with feederism, I found the stalker girl ai is so obssessed with you shes willing to get really fat with little prompt, so long as you get her to realize thats what you are into