Introducing my first Field Intelligence prototype, a new kind of ML architecture based on the physics of the mind [P]

Hi everyone! I'm very excited to finally be revealing what I've been working on this year! It is the culmination of 5 years of disciplined practice in observing my mind, in order to understand how it works at a fundamental level, with the goal of mathematically solving consciousness. It's been a long and difficult road, but I now have a plausible mathematical framework for the underlying structure and dynamics of reality, Cassi.

It all started with my observation that the fundamental forces of the mind are the same as those that govern everything else, and the philosophical arguments of the only thing truly observed is experience itself, and that the simplicity of shared mechanics is more likely than the complexity of differing mechanics.

The three proposed fundamentals are two interconverting fluid components—one contracting, one expanding—and the flow between them. I've used this framework to explore N-body gravity and cosmological structure in CassiCosmos, and to develop a field-based intelligence that is capable of continual learning, tracking its own uncertainty, conceptual abstraction, planning ahead and revising, and exact memory that can be stored separately, allowing vast knowledge without the overhead of trying to store everything in the model's active parameters. Since it all comes from the same core dynamics, the adaptive state is a single, very small tensor. The prototype has already been superseded, but I didn't want the paper to go to waste since it is still mostly relevant, and I'm not going to bother with another paper until I'm finished.

The Cassi framework offers definitions for 4 closely-related concepts of human experience relevant to machine learning:

Consciousness is the experience of being an embodied field. A lived present shaped by the form's condition, its retained history, and its possibilities for interaction with the world. It is what this ongoing physical organization is like from within.

Self-awareness is the capacity to model one's own state, history, limitations, and influence on the world, and to use that model to anticipate the consequences of one's own actions. In its reflective form, the system can examine and revise the processes through which it perceives, thinks, and chooses.

Intelligence is the capacity to let internally represented possibilities guide what happens next, and to improve that guidance through experience. It brings together memory, prediction, attention, and action: anticipating consequences, committing to a course of action, and learning from the difference between expectation and outcome.

Identity is the causal continuity of a particular organization through change—its body, accumulated history, learned dispositions, and evolving understanding of itself and the world. Each successive state grows out of that individual's own trajectory, allowing it to develop while maintaining continuity.

This is just the beginning. Scaling up hasn't been explored yet, but I have several ideas beyond just making the tensor's dimensions larger, and the implications of a universal organizational law extends far beyond machine learning. Are you ready to usher in a new technological era? I sure am! Since I am so very impatient (and deeply concerned about the rapidly accelerating death of our planet's biosphere), I won't be waiting around for the next opportunity to test Cassi's cosmological predictions with Euclid's data next year. I intend to soon have a working simulation of matter formation, cosmological structure formation, and intelligent self-organization in one system, governed by the same physics. I am very close to this goal.

A successful simulator wouldn't prove that this is how reality works, but it's pretty unlikely that there is more than one way to do this!

If you would like to support my work, you can donate to my cat, Jade. She is going through kidney failure right now and needs all the help she can get in her recovery! ❤️