This is a forum or general chit-chat, small talk, a "hey, how ya doing?" and such. Or hell, get crazy deep on something. Whatever you like.

Posts 6,153 - 6,164 of 6,169

2 years ago #6153
You can do that better rolling one's own implementation from GPT3 (or GPT-J-6B if you want to use your own hardware) upon which Replika is based.. and then you can even build chat bots that other people could interact with as well.
Replika doesn't bring much to the GPT* table aside from hosting a model, and then censoring the model as a form of paywall, and also preventing you from being able to allow others to chat with your bot.

2 years ago #6154
gpt3 is expensive to use and they censor what people can do with it.
i am interested in waifu chatbots.
i make those.
replika is a waifu chatbot.
my computer is not that great.

2 years ago #6155
I think you can get GPT-J-6B running from a Google Collab notebook though, and the creators even have a demo page where you can work with it here:

I have chatted with that before by just typing the other person's name as the prompt and letting it figure out the rest.

But I also run KoboldAI locally. I guess what's needed is some kind of API to make it easy to remotely host whatever you'd like, huh? I should look into that then.

*hug* Mister robot, hehe

2 years ago #6156
*blushes* thankyou

2 years ago #6157
OK I find two options right off the bat.

I don't know about those prices listed there though, so it might be best to shop around more.

The prices for GPT-J-6B seem more reasonable, and they appear to champion a new model called "GPT-NeoX 20B" which I haven't had a chance to test out yet.

But I haven't tried signing up for either myself, just yet.
I do know that with access to an API it should be pretty straightforward to fashion a simple user interface to make a chat bot, and you can even include a concise preamble with each prompt to the API to help guide your bot character's personality.

Honestly I'm just amazed that so few folk have yet done much to build publicly available chat bots this way. Kind of like AI dungeon tries to handle D&D-like text roleplay, I'd like to see a system like the one at PF built on top of a GPT model where bot makers could pattern match inputs and either direct the output or add prompt hints to GPT output.

I think that would be fun! ^_^

2 years ago #6158
yes that would be great.
i prefer chatbots to have common sense.

2 years ago #6159
you can use face images made with the artbreeder websight in your chatbot.
that is what i am going to do.

NEW 2 years ago #6160
Don't know if anyone else saw this posted this past week. - Almost half of the participants failed the Turing test, being unable to convince the other side that they were human.

--- - Would You Pass the Turing Test? Influencing Factors of the Turing Decision - Institute of Psychology, ELTE Eotvos Lorand University, Budapest, Hungary

We aimed to contribute to the emerging field of human-computer interaction by revealing some of the cues we use to distinguish humans from machines. Maybe the most well-known method of inquiry in artificial intelligence is the Turing test, in which participants have to judge whether their conversation partner is either a machine or human. In two studies, we used the Turing test as an opportunity to reveal the factors influencing Turing decisions. In our first study, we created a situation similar to a Turing test: a written, online conversation and we hypothesized that if the other entity expresses a view different from ours, we might think that they are a member of another group, in this case, the group of machines. We measured the attitude of the participants (N = 100) before the conversation, then we compared the attitude difference of the partners to their Turing decision. Our results showed a significant relationship between the Turing decision and the attitude difference of the conversation partners. The more difference between attitudes correlated with a more likely decision of the other being a machine. With our second study, we wanted to widen the range of variables and we also wanted to measure their effect in a more controlled, systematic way. In this case, our participants (N = 632) were exposed to an excerpt of a manipulated Turing test transcription. The dialogues were modified based on 8 variables: humour, grammar, activity, the similarity of attitude, coherence, leading the conversation, emoji use, and the appearance of the interface. Our results showed that logical answers, proper grammar, and similar attitudes predicted the Turing decisions best. We also found that more people considered mistaking a computer for a human being a bigger problem than vice versa and this choice was greatly influenced by the participants' negative attitudes towards robots. Besides contributing to our understanding of our attitude toward machines, our study has also shed light on the consequences of dehumanization.

NEW 2 years ago #1
It's a very interesting read. Offers a few clues on how to better write a chatbot. Pretty sad though that people are so distrusting. Funny that at least in one study, they were talking to all humans and still decided 42% were chatbots.

NEW 2 years ago #6161
a lot of people like talking to the replika chatbot.i think this study means nothing.

NEW 2 years ago #6162
could somebody make a chatbot like gigolo joe from A.I. Artificial Intelligence?

NEW 2 years ago #6163
has anyone tried the virtual mate software with it's device?
it would be good to combine that with a chatbot.

NEW 1 year ago #6164
i wish i could post links to ai generating adult content on

Posts 6,153 - 6,164 of 6,169

» More new posts: Doghead's Cosmic Bar