You're pledging to donate if the project hits its minimum goal and gets approved. If not, your funds will be returned.
The aim is to give AI agents control over single nations inside the simple kingdom sandbox of WorldBox. We want to see how AI agents act when they own one nation that has its religion, culture and way of life and when those AI agents face other AI agents who own the same nation.
To study how AI agents behave when they have visible control of a complex fake society.
The simplest success result is to make a mod that can run one or more AI agents and give data.
Project Goals
.Building a working setup that allows AI agents to control kingdoms in a sandbox (using existing tools and Mods)
.Run a small number of simulations (target: 10–20 runs)
.Collect structured data on AI choices, diplomacy + military decisions/results, and long term outcomes
.Produce a public report with initial findings and justifications on either scaling up the study or ending it outright
HOW
.Adapt existing open tools (WorldBox-MCP, LLM agent loop) rather than building from scratch
.Bind individual AIs like Claude, Grok, ChatGPT, etc. to a single kingdom with full control and action space
.Analyze both quantitative metrics (survival, expansion, wealth, wars) and qualitative reasoning patterns
.Keep the initial pilot deliberately small so results can be inspected and improved
.Hire a software developer: $9,000. This covers bridge setup, mod configuration, a multi-model agent loop and a logging layer for 120 hours at $75 per hour.
.Materials: $700.
.Data analyst: $3,200. This funds dataset creation, number and reasoning analysis and conclusions.
.Mods: $50.
.LLM API costs: $700.
.Margin for error: $2,050.
.Total: $16,450.
I am a researcher who wants to find practical ways to study AI agents, their thoughts and actions in a made-up complex environment. This is a project and I am now building the technical and analysis help that is needed.
I have no published record of -agent experiments but I have spent a lot of time learning about tools like WorldBox-MCP, generative agents and AI-managed kingdom mods and I have made a plan to cut costs while keeping the research useful
Most likely causes of failure:
.Technical friction: WorldBox modeling/agent integration takes longer or is less stable than expected.
.Weak or repetitive AI behavior: if the agents fall into repetitive actions and decisions.
.Insufficient runs: not being able to make sufficient test runs and gather enough data
Outcomes if it fails:
.The technical setup and logging pipeline will still be documented and published.
.Even negative results will give us insights on we're to improve or the short comings of the experiment
.The research may not generate enough value to justify its continuation.
I have raised around $1000 from the BOLA AHMED TINUBU STUDENT ASSOCIATION (BATSA) where I serve as a political science researcher and I have also applied to other sources, in order to raise funding for the research.
There are no bids on this project.