How to use AI embeddings and semantic spaces for market research

Large language models don't just generate text. They have absorbed an enormous amount of human culture, and hidden inside their embedding models is an implicit map of how concepts, brands, consumer opinions, and ideas relate to each other. This talk presents an original methodology and a visual exploration tool created to read that map: with a little geometry, we can navigate embedding space to surface insights on brand positioning, market perception, and broader cultural associations. The result: a new way to interrogate how AI models represent our world, useful for both cultural and market research.