No Way Out Agentic AI Thinks Like Boyd: The OODA Upgrade LLMs Can’t Touch
20 snips
Oct 29, 2025 Ben Ford, a Royal Marine Commando turned software engineer with expertise in Haskell and category theory, dives deep into agentic AI and the myth of the OODA loop. He argues that traditional AI models are static and inefficient, while active inference offers a more dynamic and energy-efficient approach. Ford explains how real-time model updates allow for continuous destruction and creation, substantially enhancing decision-making. With insights on category theory and its relevance, he challenges conventional thinking about knowledge acquisition and the limitations of current LLMs.
AI Snips
Chapters
Books
Transcript
Episode notes
Orientation Is The Whole Loop
- Orientation isn't a single phase; it structures and contains the whole OODA loop fractally.
- Ben Ford links neuroscience and category theory to show orientation continuously reshapes observation, decision, and action.
Create And Destroy Mental Models Each Cycle
- Destruction and creation are dual processes that reshape internal models each cycle.
- Ford frames unfolding as generating possibilities and folding as collapsing them into actionable orientation.
Unfolding As Overlapping Ripples
- Unfolding circumstances are overlapping ripples from many agents that orientation must reverse-engineer.
- Ford uses functional-programming 'unfold' to describe branching possibilities few of which are followed.
