10B Model: People Not Parameters
Modeling Needs
What does everyone in the world need? That's the goal of our 10B model, an embedding model for human needs.
Overview
Our research combines a top down approach: countries, cities, datasets; and a bottom up simulation strategy: human behavior and online activity. Where they meet is an embedding model that captures the human needs of a population. The supervised trajectory aims to predict hold out datasets of sociodemographic characteristics. The unsupervised component will then seek to reveal the hidden dynamics of the macro space.
Over time we are aiming for a system that will be of use to people's personal agents as they strive to build community and navigate the AI economic transition. People who are asking, "How can I connect with my neighbors?", "What economic value can I bring to my community?", or "How can I effectively advocate for change?"