logo80lv
Articlesclick_arrow
Professional Services
Research
Talentsclick_arrow
Events
Workshops
Aboutclick_arrow
Order outsourcing
Advertiseplayer
profile_loginLogIn

AI Simulation Platform Where Characters Make Their Own Decisions

Thistle Gulch is a Multi-Agent Gym Environment powered by Skill to Action Generation for Agents,

Why do people set up ant farms? I can't be sure as I'm not a fan of the creatures, but I think watching them build their lives and make decisions is part of the appeal. What if you could do the same with game characters? Smart NPCs are what any player and game developer wants to see, but we're still far from the point where we can talk to characters freely. However, some steps have been made in this direction with SAGA – Skill to Action Generation for Agents, a "generative AI framework that steers AI Agents toward successful goals through Actions" created by the Fable studio.

SAGA makes agents "think" for themselves and act based on their circumstances. It is inspired by NVIDIA's Voyager as well as the Generative Agents: Interactive Simulacra of Human Behavior research, which analyzed a small virtual town populated by AI agents who do the same we do every day: they wake up, cook breakfast, head to work, and celebrate birthdays. 

Image credit: Fable

"With SAGA, Agents first tell SAGA contextual metadata about themselves and their world via a companion simulation: Who they are; What they know; What “Skills” they have; And what their goals are. Then, when an Agent is deciding what to do next, SAGA generates a set of “Actions” that best serve the Agent’s goals in that moment. These Action options are then scored and returned to the simulation in order to direct the Agent. This process repeats each time the Agent is deciding its next action and can be scaled to multiple agents running simultaneously."

Agents implement skills and thus perform various actions.

To give them a place to do this, the developers from Fable set up Thistle Gulch, a simulation platform in the form of a fictional 1800s Wild West town featuring over 15 characters with unique backstories and a murder scenario. A local sheriff has to use his AI brains to find the murderer with Actions. 

Image credit: Fable

"In order to generate Actions, the simulation needs to provide SAGA with relevant details we call “Meta-Memories”. These are the sum of all the relevant metadata surrounding the simulated world and its Agents. Some of this information will only be available to specific characters, like memories and observations. Other information is shared across characters like the locations, interactable objects, and summaries of other characters.  

Meta-Memories are loaded just before the simulation starts or are streamed in while the simulation is running, as events unfold. This includes details about the activities other characters are performing, their conversations, and new memories that are formed as the simulation progresses."

Just like in real life, there are multiple ways to solve the case, and it's fascinating to watch little people live their lives. You can find more details on that in this blog post.

Image credit: Fable

Fable says its platform is designed for creators, researchers, and AI enthusiasts, offering "unprecedented control over the narrative and character interactions through a powerful Python API. This allows for deep customization of AI decision-making processes and conversation generation."

You can join the research in beta to help its development and learn more about AI behaviors. The team is looking for storytellers, writers, AI enthusiasts, researchers, educators, students, game designers, developers, and other interested parties.

Apply here and join our 80 Level Talent platform and our Telegram channel, follow us on InstagramTwitter, and LinkedIn, where we share breakdowns, the latest news, awesome artworks, and more.

Join discussion

Comments 0

    You might also like

    We need your consent

    We use cookies on this website to make your browsing experience better. By using the site you agree to our use of cookies.Learn more