This is an AI scenario generator that can create scenarios based on conflict of NPCs of different ideologies in a fictional or reality-based setting.
This project is a piece of a larger game that I'm developing that attempts to map user's political ideologies based on their answers to silly questions revolving around squirrel land rights, squirrel healthcare, and similar, completely squirrel related, political questions.
There is a general anger and malaise rising in the global population. People are fed-up with legacy systems of governance that have ballooned in size and budget to support violent bureaucracies while ignoring pressing problems and crises. Resources are unfairly distributed, people are sick and hungry, prices are rising, and every day there is a new story detailing some failure of government or outrageous incompetency. It seems that waging war, or threatening to do so, is one of the few things that governments are still able to do.
While we continue to innovate in all sectors of the economy while innovation in systems of governance have stalled. We leave governance to more learned, legal minds, and decry the ways government has failed us, without proposing or pursuing change on a more systemic level.
Squirrly is an attempt to engage people in a process of governance experimentation by understanding better what it is they think government should actually do, and allowing them to build one from scratch.
Users will engage in a generative storytelling game, where decisions are voted on, ideological impact of decisions are logged, and a squirrel government is formed to protect the squirrel nation from global catastrophe, internal strife, and attack by birds to hopefully arrive at ideas for better systems of governance, and more broadly, spark conversation about governance innovation. A conversation we desperately need to have if we are to survive all the natural and unnatural horrors that will confront us as we move forward in time.
I was working on the scenario generation engine for this hackathon. With the help of my brother, we built a querying and classification engine using the OpenAI API and langchain to string together prompts and produce a data object that can be written to chain and pushed to users on a daily basis.