Thoughtforms Life Karl Friston, Adam Goldstein, and Michael Levin Discuss Active Inference and Algorithms
Feb 4, 2024
Join Karl Friston, a leading theoretical neuroscientist known for his work on active inference, alongside Adam Goldstein and Michael Levin, as they delve into the fascinating intersection of algorithms and self-organization. They explore how simple sorting algorithms can exhibit unexpected behaviors and emergent properties. Friston connects these concepts to biological systems through free energy principles, while the trio discusses the implications for social dynamics, revealing insights into clustering and the tension between mechanics and teleology. It’s a thought-provoking dive into the algorithms that shape our world!
AI Snips
Chapters
Transcript
Episode notes
Sorting Algorithms As Minimal Minds
- Michael Levin framed simple sorting algorithms as minimal models to study emergent, distributed intelligence under local constraints.
- Visualizing sorting as navigation with broken cells revealed surprising behaviors like delayed gratification and clustering.
Free Energy View Of Sorting
- Karl Friston linked the sorting behavior to self-organization and self-organizing maps, suggesting Markov random fields as a useful formalism.
- He recommended finding the Lagrangian or free-energy functional that the system implicitly minimizes.
Digits As Markov Blankets
- Friston argued each digit can be treated as a Markov blanket exchanging local sensory and active states, enabling an active inference interpretation.
- He said similarity of algotype can reduce surprise, explaining clustering as a free-energy minimizing outcome.