Vision
AI now helps decide what we read and buy, how buildings and cities run, and how machines move among us. Most of it learned from text and data: what people wrote down about the world. Living Worlds lets AI learn from experience of the world instead, by living through it: lives with their choices and consequences, and places with their sound, light and movement, in simulated worlds that are checked against the real one.
Our position
AI trusted with real-world decisions has to learn from experience of the world, and be measured against the world before it is trusted.
Text holds what people learned, second hand. In Living Worlds an AI lives through months and years: choices, consequences and memory, in worlds built from measured rates and measured spaces.
Every model is registered and scored on real data it has not seen, and used within the range where it was shown to hold. Where a rate is still a placeholder, the study says so. The results are public in Research.
Experience does not decide what an AI should want. Anything that acts on the physical world or spends money waits for a person, and some things are not learned by experiencing them at all.
What we simulate
People and how they live: how they answer, form families and change over a lifetime. Places and what moves through them: sound, light, crowds and power, with water next. Each new domain arrives the same way: built from measurements, checked on data it has not seen, and published.