Install Steam
login
|
language
简体中文 (Simplified Chinese)
繁體中文 (Traditional Chinese)
日本語 (Japanese)
한국어 (Korean)
ไทย (Thai)
Български (Bulgarian)
Čeština (Czech)
Dansk (Danish)
Deutsch (German)
Español - España (Spanish - Spain)
Español - Latinoamérica (Spanish - Latin America)
Ελληνικά (Greek)
Français (French)
Italiano (Italian)
Bahasa Indonesia (Indonesian)
Magyar (Hungarian)
Nederlands (Dutch)
Norsk (Norwegian)
Polski (Polish)
Português (Portuguese - Portugal)
Português - Brasil (Portuguese - Brazil)
Română (Romanian)
Русский (Russian)
Suomi (Finnish)
Svenska (Swedish)
Türkçe (Turkish)
Tiếng Việt (Vietnamese)
Українська (Ukrainian)
Report a translation problem
What they need to do is offer the AI model as a service or API app.
They could even monetise the online service where everyone gets the basic crap 1B model for free. If you want a 22B+ model you can pay a monthly fee.
That or just download the API app and install it on another machine and use its GPU and use it for free that way.
I really hope they don't try to turn that into a service "just to provide" autonomy. Unless it'll start increasing normal conversation instead of just clicking popups to "level up" traits/skills etc and actually allow a Zoi to listen to that model. I don't know much on game dev, but each Zoi's going to eat alot of request alone via API.
From my experience, certain actions like Cooking are just blocked and will never be done autonomously. In a sense, this applies to also non-SmartZoi AI.
They're inter-related from my experience and that of a few others. Some prompts actually influence the autonomy and actions, like, if I say "you are a clean freak", my Zoi started doing more cleaning actions without changing any preferences.
But you're right, maybe my feedback wasn't on the SmartZoi itself, but it does make it feel weaker when my Zoi goes to play games and says "I should go whip up something" and never starts to cook
Think I'm going to uninstall and check back in 6 months or so. Maybe they'll have that fixed because right now, unless you really love micro managing every thing your zoi does, it's just not fun. I like full autonomy.
"You're a best-selling author and literary genius. Your expertise spans crafting gripping plots, creating vivid characters, and writing evocative prose. Your answers are imaginative, clear, and designed to spark creativity in users' writing projects."
if you want them to be better at writing (haven't tested this prompt yet)
In the end I disable it again because my graphic dropped from ultra to high and I prefer to play without autonomy. No interest in other Zoi’s and how they behave. Maybe they drop this smartZOI and rather work on generally improving the autonomy without the help from AI.
This is going to be a problem for awhile. Even the latest modern GPU's don't really have the VRAM or spare compute to run AAA titles with a decent AI model in the background. It might be 5 years before we start seeing cards capable of managing that comfortably.
I'm writing a Python front end for LLM models to generate and DM full text games with images and 2D maps and the biggest issue I'm running into if I ever get around to finishing it and release to open source is most people won't have the hardware requirements to run the 32B models I'm pairing with it to get it to run well.
A work around solution for now would be offering the AI model as an online service or and offline API app that can be run on another machine where its got the full GPU resources available. The smaller models are improving but anything under 8B is pretty bad still.