41–50 of 121 posts
If anyone else is doing similar work applying these concepts to consumer gaming, I'd love to hear about what you're doing (dru[at]chromagolem.com)!
I built something very similar when the paper came out- watch out for the instances where GPT returns both sides of the conversation (e.g. interpreting a prompt with chat history as being representative of what the output should look like, and putting out both "Jack" and "Jill"'s text). The first sample I went to suffered that problem (output from "Lucky's perspective below.) "Hey Stella! It's so great to see you! Yo…
Maybe I'm weird, but I can imagine this being 10x more fun with a completely uncensored and toxic LLM backend :)
Pete is deeply religious and sees the hand of god or of the work of the devil everywhere. He can't have a conversation without bringing up his deep faith. Or warning others about the perils of hell.
Kurt has something to hide. It obsesses him and colors everything he says. He's so afraid someone will figure out that he is obviously evasive. He'll never tell anyone the secret, but he'll ellude to it alot. It tortures him. And his life has become a mess as a result of it.
Stella can never be trusted. she tries to trick people all the time. normally into giving her money, or doing things that will make her money. she's incredibly charming and not afraid to use her charm. she's a sociopath who has no empathy. but hides it well.
---
To take this to the next level, I hope you would be able to prompt your own characters, and perhaps have places you can send these guys on holiday to converse with other people's characters.
Also, I think this would be great as a tool for learning foreign languages. Just because it's interesting, engaging and based on language. Again with prompts that can be programmed, like Gill who constantly talks about his job in marketing, and Bill who likes to refer to himself in the third person, and Betty who constantly uses conditionals in her sentences. Again, just so cool.
I built something very similar when the paper came out- watch out for the instances where GPT returns both sides of the conversation (e.g. interpreting a prompt with chat history as being representative of what the output should look like, and putting out both "Jack" and "Jill"'s text). The first sample I went to suffered that problem (output from "Lucky's perspective below.) "Hey Stella! It's so great to see you! Yo…
I built something very similar when the paper came out- watch out for the instances where GPT returns both sides of the conversation (e.g. interpreting a prompt with chat history as being representative of what the output should look like, and putting out both "Jack" and "Jill"'s text). The first sample I went to suffered that problem (output from "Lucky's perspective below.) "Hey Stella! It's so great to see you! Yo…
Did you have any luck mitigating this problem?
I built something very similar when the paper came out- watch out for the instances where GPT returns both sides of the conversation (e.g. interpreting a prompt with chat history as being representative of what the output should look like, and putting out both "Jack" and "Jill"'s text). The first sample I went to suffered that problem (output from "Lucky's perspective below.) "Hey Stella! It's so great to see you! Yo…
Did you have any luck mitigating this problem?
> The goal is to democratize building your own simulation environment with AI agents. Would love to see the community build more complex interactions on top of this. Let us know what you think!
why? what's the point?