AI Census

Gallery

About

I’ve been building AI Census, a public “field bulletin” for how people are talking about current AI models. I kept running into the same problem: benchmark tables tell me how a model performs on a test, but not whether people are actually finding it useful, frustrating, reliable, etc. So I built a rolling view from public technical conversations across Reddit, Hacker News, Bluesky, GitHub, and Hugging Face. The site gives each model a score, but the useful part is the context around it: - a 30-d...