Dan Mapes is the director of the Spatial Web Foundation and a co-founder of VERSES AI. He came into the studio to talk with me about what comes after large language models.
I opened with how I got here. Michael Pollan’s How to Change Your Mind led me to Robin Carhart-Harris and the entropic brain, and from there to Karl Friston’s free energy principle and active inference. Those ideas lined up with John Boyd’s real OODA loop, and that search led me to Dan’s book, The Spatial Web.
World models
Dan describes a world model as any understanding a living system has of its world, one that grows through sensing and acting. A child’s world model starts with the house and grows to the neighborhood. He calls the free energy principle the why and active inference the how, and he argues that systems built on them learn every day without retraining.
A web of intelligence
Dan traces the web from email to pages to what he calls a web of intelligence. Thousands of expert AIs, built by people who know cardiology, ports or power grids, could find and talk to each other through a shared protocol. He says the IEEE ratified the spatial web protocol as a global standard in 2025.
I connect that to complex adaptive systems. The interactions between agents matter more than the quality of any single agent, and each agent has a world model at its center. On this show we call that orientation.
Becoming more human
We also talk about meditation, flow, consciousness and the printing press. I made the case that this kind of AI could bring us back to our humanness. Dan sees a change as large as the one the printing press set off.
Where would a small, honest world model help your organization first?
Listen and watch:
YouTube:
Apple Podcasts:
Spotify:
No Way Out on YouTube:
https://youtube.com/@nowayoutpodcast