Excellent presentation. Homeostasis has long interested me and that functions on a similar (albeit more complex) way, simple mechanisms based on simple rules producing an emergent 'system' that addresses a whole series of problems.

❤-it

We need more understanding of swarm intelligence, and what models result in emerging phenomena on network level that is not apparent on local level. In machine learning, I believe this would bring us to a new level of complexity/intelligence - social interactions bring forward a faster / "better in some ways" knowledge, our current models (I mostly hint at Reinforcement Learning, not LLMs, but LLM agents would be the same) do not have common sense which may come from no "common knowledge" if there is no one to share the wisdom with. I think if humans were not social creatures, we would not have much common sense either, that emerges from society and adapting to society. To my understanding, we do not have realistic, general frameworks for having compact agents whose behavior collectively emerge to something great. (Of course, we have particle swarm optimization and models for bird flocks, but none seem to really "emerge without a designed rule".) And because these agent models are still not compact enough to not be computationally expensive, especially when running a large network of agents (hard to paralelize, as agents interact), not much research goes into swarm and social intelligence in computer science...