Intelligence as an omnipresent phenomenon

 Is it possible that "intelligence" is an omnipresent phenomenon? Or even _the_ substrate of existence?

We already know that human-alike intelligence could be trained by a couple hundred lines of code (and lots of data and compute).

In the physical universe, data and compute can be "free" (if harnessed correctly). Trillions of molecules (as both data and computational things) are present in every speck of matter in our world.

We also know that language is limiting. Multimodal models seem to suggest that instead of using language as protocol, if we just feed neural networks vectorized stuff and give them some time to "learn" and adopt to the language, we don't need human-understandable language for things to work.

In fact, it seems to me that language is a limiting factor in such cases. Instead of vectorized representations of "meaning", language forces these vectors into a single dimension and small number (~10-20?) of bits for each word. We struggle to communicate with words, since they encode meaning poorly. 

So it's reasonable to hypothesize that if we plug two objects that emit signals together, they will be able to learn and communicate automagically. We won't understand the protocol since we always try to compress the vast amounts of signal into our language, but apparently this just works in nature (even across species).

The use of language is probably a mixed blessing - on the one hand it is a clear and physical manifestation of a simple protocol to convey meaning. On the other, the limitation of bandwidth (~20 bits per second?) severely retards the potential of thought and intelligence. The fact that we believe language is the pinnacle of human intelligence shows how naïve we are. And it probably explains the parable of the Tower of Babel as well....

----

So there are a couple weird ideas:

1. Does "computation" even need "data" to become intelligent? It depends on whether there is such a thing as universal intelligence? Consider that the structure to simulate worlds can be minimal (see eg. Mandelbrot, game of life, and even laws of physics), and these worlds can generate vast amounts of "data", it seems that even if computation systems are isolated to places with very little entropy or noise, the systems will eventually be able to achieve some kind of intelligence by simulating worlds and learning from them. This seems like a candidate for a theory of cosmology (i.e. we are just in one of such simulations) based on our latest understanding of the power of computational networks and generative systems.

2. if computational structures of the right type somehow exist in the universe, then intelligence is ubiquitous, omnipresent (and perhaps to a large extent omniscient). The "problem" is that, just as we do not understand the weights of LLMs, the meaning of the vectors of word embeddings, etc., we do not understand this universal "field" of intelligence, and attribute it as noise.