We recorded this conversation with Dr. Mahault Albarracin on the 49th anniversary of John Boyd’s Destruction and Creation. Mahault is Director of Research Strategy at VERSES AI. Mark “Moose” McGrath and I asked her what endless scaling misses.
Mahault came to active inference from the social sciences. She describes it as a framework that models intelligence on natural cognitive systems. That echoes Boyd’s point about breaking down old mental models to build new ones that fit.
Context and embodiment
We look at the parallels between Karl Friston’s active inference and Boyd’s real OODA loop: prediction, orientation in a complex environment, and keeping tactical actions in harmony with changing strategic intent. Mahault explains why current AI struggles with tasks people find easy. It lacks a body in space and time, and it misses how context changes meaning.
Doom and kinship
Mahault challenges the common AI doom stories. She traces them to biases in defense funding, old hierarchies and an adversarial view of the world. She argues that alignment depends on giving AI the capacity for empathy, so it can see people as kin.
Harmony between agents
We cover the spatial web, harmony in systems of many agents, and how resonance between agents, from birdsong to human closeness, might inform how people and AI work together. Orientation sits at the center of Boyd’s loop, and Mahault’s work asks what an artificial agent would need to orient well.
What would it take for you to trust an AI teammate?
Listen and watch:
YouTube:
Apple Podcasts:
Spotify:
No Way Out on YouTube:
https://youtube.com/@nowayoutpodcast